WPS7315 Policy Research Working Paper 7315 Product Standards and Firms’ Export Decisions Ana M. Fernandes Esteban Ferro John S. Wilson Development Research Group Trade and International Integration Team June 2015 Policy Research Working Paper 7315 Abstract The paper estimates the effect of product standards on firms’ deters exporting firms from entering new markets and leads export decisions using two novel datasets. The first covers to higher exit rates from those markets. Moreover, firm all exporting firms in 42 developing countries. The second characteristics mediate the effect of product standards on covers pesticide standards for 243 agricultural and food firms’ export decisions. Smaller exporters are more neg- products in 63 importing countries over 2006–12. The atively affected in their market entry and exit decisions analysis shows that product standards significantly affect by the relative stringency of standards than larger export- foreign market access. More restrictive standards in the ers. Positive network effects of exporters from the same importing country, relative to the exporting country, lower country may help reduce the burden of importing coun- firms’ probability of exporting as well as their export values tries’ standards on firms’ decisions to enter new markets. and quantities. The relative restrictiveness of standards also This paper is a product of the Trade and International Integration Team, Development Research Group. 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://econ.worldbank.org. The authors may be contacted at afernandes@worldbank.org and eferro@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 Product Standards and Firms’ Export Decisions    Ana M. Fernandesa Esteban Ferrob John S. Wilsonc The World Bank The World Bank The Center for Global Enterprise Keywords: Exporter dynamics, Entry, Exit, Intensive margin, Extensive margin, Non-tariff measures, Product standards. JEL Classification codes: F14, Q17, O13, L15.                                                               We are grateful to Jose-Daniel Reyes as well as participants at the 2014 European Trade Study Group conference and the World Bank DECTI brown bag for comments. Research for this paper has been supported in part by the World Bank’s Multidonor Trust Fund for Trade and Development and through the Strategic Research Partnership on Economic Development. We also acknowledge the generous financial support from the World Bank research support budget and the Knowledge for Change Program (KCP), a trust funded partnership in support of research and data collection on poverty reduction and sustainable development housed in the office of the Chief Economist of the World Bank. The findings expressed in this paper are those of the authors and do not necessarily represent the views of the World Bank. a Senior Economist, Trade and International Integration Unit, Development Research Group, World Bank (afernandes@worldbank.org). b Economic Consultant, Trade and International Integration Unit, Development Research Group, World Bank (eferro@worldbank.org). c Fellow, The Center for Global Enterprise, New York, New York. (jwilson@thecge.net).  1. Introduction While the substantial declines in tariffs over recent decades under World Trade Organization (WTO) negotiations and in the context of preferential trade agreements have fostered growth in world trade, countries have been tempted to resort to alternative ways to protect their domestic markets. The growing use of non-tariff measures (NTMs), generally in the form of import regulations, by both developed and developing countries has resulted in a trade policy environment that is less transparent and less well understood.1 Anecdotal evidence suggests that NTMs restrict market access, particularly for smaller exporters in developing countries. But the effects of NTMs on trade are difficult to assess due to the breadth of regulations covered as well as their non-measurability. In this paper, we provide rigorous econometric evidence on the effect of product standards—a type of NTM—on trade using firm-level data. We estimate the effect of standards on pesticide residue limits for agricultural and food products imposed by importing countries on firms’ decisions to export, enter or exit a product-destination market, as well as their export values and quantities. To do so we combine two novel datasets, one covering all exporting firms in 42 developing countries and one covering pesticide standards for 243 agricultural products in 63 importing countries over the 2006-2012 period. Agro-food products are an important component of the export portfolio of the developing countries in our sample, accounting on average for 20 percent of their total exports, and play a critical role for the development of poor rural areas in most developing countries.2 Our main findings are as follows. First, product standards affect foreign market access in that more restrictive product standards in the importing country relative to the exporting country reduce significantly firms’ probability of exporting as well as their export values and quantities. Second, our evidence shows that the relative restrictiveness of importing countries’ standards deters exporting firms from entering new product-destination markets and leads to higher exit rates from product-destination markets (once firm size is controlled for). Third, firm characteristics mediate the effect of product standards on firms’ export decisions. Smaller                                                              1 NTMs are a key topic of negotiation under new trade agreements such as the ongoing Transatlantic Trade and Investment Partnership (TTIP) negotiations between the United States (US) and European Union (EU). Since tariffs in the EU and US are low, any substantial impact of such an agreement on trade flows for signatories and third markets will be driven by changes in NTMs.  2 The average masks heterogeneity across countries in the share that agro-food products represent in total exports, ranging from 50-60 percent for Kenya, Nicaragua, and Uganda to 5 percent or less for Bangladesh, Botswana, Cambodia, and Mexico.  2    exporters are more negatively affected in their market entry and exit decisions by the relative stringency of standards than larger exporters. The presence of firms from the same country exporting the same product to a given destination – network effects – reduces the negative effect of the relative restrictiveness of destination country standards on firms’ decisions to enter a new destination market. The Sanitary and Phyto-Sanitary (SPS) standards considered in this paper are Maximum Residue Levels (MRLs) which restrict the maximum levels of residues from pesticides legally permitted on unprocessed food.3 MRLs are mandatory regulations which condition market access in order to ensure that domestically-produced and imported unprocessed food is safe to eat.4 Countries are quite heterogeneous in the products they regulate, the pesticides they regulate for each product, and the MRL they allow for a given product-pesticide pair.5 In order to meet required MRLs, producers need to avoid using certain pesticides and determine the correct pre- harvest intervals, which often requires the use of more expensive inputs or specialized human capital. Non-compliance with an MRL can lead exporters to lose the full shipment value and be subject to additional monitoring and testing until multiple shipments successfully cross the border.6 The contribution of our study is three-fold. First, ours is the first study that examines the relationship between explicit measures of product standards and firms’ export decisions for a large set of developing countries. Second, in contrast to most previous studies which rely on a count of standards, our use of MRLs for pesticides allows us to construct index measures that quantify the absolute stringency of these standards. Third, our study focuses on differences in the stringency of standards in the importing country versus the exporting (home) country. If the MRL is stricter in the importing country than in the exporting country, then firms may need to incur further production costs to meet the stricter MRL in the destination market. If the opposite                                                              3 Residues from pesticides are very small traces of pesticide that sometimes remain on treated crops.   4 Once pesticides are demonstrated to be safe for consumers, MRLs are set by independent scientists, based on rigorous evaluation of each pesticide. MRLs act as an indicator of the correct use of pesticides and ensure compliance with legal requirements for low residues on unprocessed food.  5  The Agreement on the Application of Sanitary and Phytosanitary Measures approved by WTO members in the 1990s embodies the principle that multilateral trade rules allow countries to adopt measures to protect human, animal, or plant life or health, provided such measures do not discriminate or are not used as disguised protectionism. That Agreement sets out the basic rules for ensuring that each country’s food safety and animal and plant health laws and regulations are transparent, scientifically defensible, and fair.  6 Non-compliant exporters also have to pay the cost of shipment along with storage fees overseas while the complaint is processed, a fee to dispose of the product or re-export it, and potentially a violation fee. At the extreme, MRL violations lead to complete import bans for specific origin countries. For example in 2002 imports of frozen spinach from China were banned by Japan due to the finding of a pesticide (Chlorpyrifos) in excess of the allowable MRL.  3    is true then the importing country’s MRL should not constitute a constraint on potential exports by firms in that exporting country. Considering the relative stringency of standards in the importing country with respect to that in the exporting country adds variability to our standards measures and thus enables us to include in all our econometric specifications a stringent set of fixed effects to control for, among others, unobserved country-industry factors (such as domestic competition) that could bias the estimated effect of standards. Moreover, in some specifications we control also for unobserved time-varying exporting country, destination country, exporting country-product heterogeneity (via fixed effects) as a way to mitigate potential biases caused by omitted variables. In the context of recent trade models with heterogeneous firms, following Melitz (2003), conforming to regulatory standards in an importing country constitutes a fixed entry cost to penetrate that market, as discussed by Bernard et al. (2011). But at the same time, conforming to regulatory standards may be part of the variable trade costs that need to be incurred every time the firm exports to that market, e.g., if more costly inputs need to be used to meet the standards. The models of Chaney (2008) and Bernard et al. (2011) examine the effects of destination- specific or product-destination-specific fixed trade costs as well as destination-specific variable trade costs on the extensive and the intensive margins of firm exports. They show that variable trade costs affect both margins of firm exports, if those costs decline each existing exporter exports more and new firms enter the export market (since the threshold productivity level above which firms have positive profits from exporting decreases).7 Fixed trade costs affect the extensive margin but they do not affect the intensive margin, i.e., the firm’s exports to that market (because existing exporters have already paid this cost). In a simpler model with a single firm making export decisions, Chen et al. (2008) allow compliance with standards to impose additional production costs on firms but also to possibly have a positive effect on demand (in terms of consumers’ willingness to pay for the products). The net effect of standards on a firm’s choice of optimal scale and export scope depends on the strength of the standards-induced increase in costs versus the strength of the standards-induced increase in demand. Overall, our findings demonstrate a negative effect of the stringency of SPS standards on the extensive margin of firm exports, namely in terms of entry into new markets, as well as on                                                              7  In these models a firm’s profitability of exports varies according to the destination market and it is more profitable for a firm to export to a market with high demand as well as low variable trade costs and low fixed costs. Therefore, only a subset of firms are able to export and that subset varies with the characteristics of the destination market.  4    the intensive margin of firm exports. Also, our findings show that beyond average effects there is important heterogeneity across exporting firms. Thus, our findings are consistent with the idea that firms face not only a single fixed cost to export but also additional fixed costs to meet foreign standards in each destination market. Only those firms that are productive enough—size being a proxy for productivity—are able to cover those costs and enter a new foreign market. From a policy point of view, SPS standards differ across countries and this heterogeneity is likely to remain in place since full global harmonization of such standards is not likely. At the multilateral level, the WTO SPS agreement has not been able to bring discipline for countries to adhere to the Codex Alimentarius standards and at the bilateral level SPS standards have only very recently begun to be discussed in the context of deep trade agreements (that include provisions on technical barriers to trade) but with little progress achieved to date. Thus, for a developing country that seeks to support its agricultural exporters in terms of their entry and survival in markets with more stringent SPS standards, the best strategy is for the government to help exporters meet foreign standards by creating the necessary testing facilities and customs clearance procedures, and by facilitating the imports of the necessary inputs to meet foreign standards. The remainder of the paper proceeds as follows. Section 2 briefly reviews the existing literature on product standards. Section 3 describes the data while Section 4 presents the empirical framework. Section 5 discusses the main results and Section 6 focuses on the robustness tests. Section 7 concludes. 2. Brief Literature Review The effect of the two major types of NTMs - Sanitary and Phyto-Sanitary (SPS) and Technical Barriers to Trade (TBT) measures which address regulatory standards on agricultural and food products and manufacturing goods - on trade is ambiguous, particularly for developing countries. On the one hand, regulatory standards can impede trade flows by explicitly banning such flows or by imposing prohibitive costs of compliance—related to upgrading production systems, acquiring special types of processing and storage equipment, and implementing quality control procedures—for firms in developing countries that can undermine the competitiveness of their exports.8 Survey evidence suggests that fixed costs of compliance with product standards                                                              8 See Henson et al. (2000) on the challenges related to product standards faced by firms in developing countries.   5    are important though recurring costs of compliance tend to be lower.9 The inability to comply with standards is costly not only for individual firms but also for a country’s reputation as it can ultimately result in trade restrictions such as import bans for specific products from specific countries.10 More broadly, regulatory standards can impede trade if used in a protectionist way, being more stringent than what scientists determine as acceptable. On the other hand, standards can act as trade facilitators by signaling that products are safe to the consumer (which is valuable under asymmetric information) and by providing the incentives for developing countries to modernize the supply chain structure in their export sectors (e.g., increasing investments in quality assurance), enabling them to improve their competitiveness while also strengthening the standards domestically. Maertens and Swinnen (2009) show that foreign standards acted as a catalyst for production upgrading in Senegal. Most studies on product standards and trade examine how aggregate trade flow are affected by technical regulations in a gravity regression framework.11 Disdier et al. (2008) show a trade-impeding effect from a higher number of SPS and TBT regulations imposed by OECD countries that significantly reduce exports by developing countries.12 Anders and Caswell (2009) find a significantly negative impact of stricter food safety standards on U.S. seafood imports especially from developing countries, while Tran et al. (2011) find a significantly negative impact of a zero tolerance for a drug residue standard on crustacean imports by Canada, the EU15, Japan, and the U. S. from major Asian exporters. In contrast, Xiong and Beghin (2010) examine the effect of the tightening of an EU SPS standard on aflaxotin in 2002 and find that it had no effect on African exports of groundnuts, which were instead hampered by domestic supply constraints. Ferro et al. (2015)—who compiled the standards restrictiveness index based on data on MRLs of pesticides imposed by importing countries that we use—show that more restrictive standards are associated with a lower probability of observing a positive trade flow but do not affect trade volumes per se. They interpret this finding as indicating that meeting stringent                                                              9 Using firm-level data from the World Bank Technical Barriers to Trade Survey for 16 developing countries, Maskus et al. (2005) show that fixed costs represent on average 425,000 U.S. dollars per firm (or 4.7 percent of value added) but the elasticity of firm variable production costs to standards and technical regulations is only in the 0.06%-0.13% range. Case study evidence for shrimp exports from Nicaragua shows that fixed costs to comply with quality and safety standards represent less than 3% of total annual exports while costs to maintain compliance represent less than 1% of that total (Cato et al., 2005).  10 For example, the EU ban on fish imports from Kenya decreased the country’s export earnings by 37% (Henson et al. 2000).  11 See Cadot and Malouche (2012) for a review of the role of standards and technical regulations for trade.  12 Other examples of trade-impeding effects are Otsuki et al. (2001a, 2001b) who show lower edible groundnuts exports by African countries resulting from tightening an SPS standard and Wilson et al. (2003) who show lower bovine meat imports in countries imposing more stringent food safety standards.  6    standards increases the fixed costs of exporting, but once firms enter the market, standards do not impact their level of exports.13 To our knowledge only three studies examine how standards affect firms’ trade patterns. The first study by Chen et al. (2008) uses cross-sectional data from a World Bank survey of firms covering compliance with technical barriers to trade and firm participation in export markets and show that different types of standards exhibit different links with intensive and extensive margins of exports. Quality standards and labeling requirements are positively correlated with firms’ average export volume across destinations and products but also with their export scope (number of destinations and products) while certification procedures are linked to a decline in export scope. A clear limitation of that study is its use of subjective survey responses by firms of whether their exports have been impacted by different types of standards. The second study by Reyes (2012) shows that the harmonization of EU electronics regulations with international standards (whose compliance is not compulsory) led to the entry of new US exporters of electronics products into the EU. Finally, the third study by Fontagné et al. (2015) examines the impact of SPS concerns raised in the WTO Committee on SPS on the export behavior of French firms using customs data. SPS concerns have a negative impact on firms’ probability of exporting and export value and lead to increases in their export prices. They also find a heterogeneous effect across firms, with the negative effect of SPS concerns being lower for larger firms exporting to multiple destinations. While very informative, the key limitation of their study is that relying on SPS concerns as a measure of importer restrictiveness raises potential endogeneity concerns, as only countries whose exports are depressed by an importer’s SPS are likely to raise a concern at the WTO and thus we expect greater SPS concerns to be linked through this reverse causality channel to a more negative impact on exports. The authors, however, argue that their data are preferable to the use of WTO notifications or traditional sources of information on the existence of a regulation. 3. Data                                                              13 All studies mentioned focus on mandatory technical regulations—as our MRLs are—but other studies focus on voluntary technical regulations. Czubala et al. (2009) show an inhibiting effect of voluntary EU standards on African exports of textiles, clothing, and footwear (except for those standards that are internationally harmonized) whereas Shepherd and Wilson (2013) show that voluntary standards in EU food and agriculture markets are trade-inhibiting for all countries and for raw or lightly processed goods, but internationally harmonized EU standards have much weaker trade effects, and in some cases are even trade- promoting.  7    3.1 Data on Standards Our policy variable measures the restrictiveness of mandatory standards imposed on agricultural products based on MRLs of pesticides allowed for agricultural products in several importing countries.14 The source for our data is the Homologa database obtained from Agrobase-Logigram, a French company that collects information on monthly changes in allowable pesticides for 63 importing countries from each country’s relevant ministry and standardizes the information in terms of language, unit, and format for the period 2006-2012. The Homologa dataset reports only the importing countries’ list of regulated pesticides. However, many countries use a ‘deferral policy’ for pesticides for which it has not set a specific limit: e,g., in the case of the EU the default MRL is 0.01 parts per million (ppm). Many countries defer to Codex Alimentarius standards, the set of international standards for food safety and consumer protection developed by the Food and Agriculture Organization and the World Health Organization. In fact, many countries, particularly developing countries, do not have a list of regulated pesticides but instead directly defer to Codex standards. For the empirical analysis we will use all the information available for each importing country, including the countries’ deferral policies as well as their default MRLs. Appendix Table A.1 provides the list of importing countries for which we have data on pesticide standards and indicates whether each country has its own set of regulations and what is each country’s deferral policy. Using Agrobase-Logigram’s Homologa data we matched 243 agricultural products to their corresponding harmonized system (HS) codes at the 6-digit level of disaggregation.15 The products covered are agricultural products belonging to HS Chapters 06-24 with the exception of HS Chapters 15 and 16 (oils and edible preparations of meat and fish).16 Table 1 lists for each importing country the number of HS 6-digit products covered by MRLs as well as the number of HS 6-digit products regulated including deferral and default policies. The number of different products regulated in total by all importing countries range from 214 in 2006 to 241 in 2012. The product coverage is very heterogeneous across countries: e.g., in 2012 pesticide limits were set by Brazil on 73 products and by the EU on 136 products. However, the product coverage of each                                                              14 Note that these standards apply both to domestic production and to imports of the products.  15 While Homologa’s product coverage is greater than 243 products, we were unable to match all products directly to an HS code.   16 We omit HS Chapters 15 and 16 from the analysis because they include animal products for which importing countries also regulate veterinary drug MRLs which are not covered by our standards dataset.   8    country is fairly constant across time.17 The importing countries with MRL regulations with the widest product coverage are the US, Canada, and Australia. However, importing countries such as Japan, Canada, and the EU which have default MRLs are the countries with the widest coverage of products once default MRLs are accounted for. The same is true when we analyze the breadth of MRL regulations regarding the number of pesticides in Appendix Table A.2. In total, the importing countries in our sample regulated 863 pesticides in 2006 whereas by 2012 this number had increased to 992. Japan, the Republic of Korea, Switzerland, and the EU have the most extensive coverage of specific pesticides. There are three challenges when working with MRL data. First, there are two dimensions of restrictiveness that need to be considered: the number of regulations per product and how strict those regulations are. Second, the heterogeneity of pesticides regulated across products and countries makes it difficult to compare how restrictive one country is relative to another. For example, the 16 pesticides regulated for oranges in the Russian Federation in 2011 might not be included among the 102 pesticides regulated for oranges in Brazil. Thus the question arises: how do we compare Russia’s and Brazil’s restrictiveness for oranges? Averaging MRLs across pesticides for each country-product pair would generate a misleading measure. Pesticides differ on their degree of toxicity and thus highly toxic pesticides always tend to have lower MRLs than less toxic pesticides. The group of pesticides a country chooses to regulate for a specific product will determine if the average MRL is high or low without providing any information on the restrictiveness of those measures relative to other countries. Consequently, it is imperative to normalize MRLs at the product-pesticide level to get a true measure of how restrictive a country’s MRL is, relative to how all other countries are regulating that same pesticide for a given product. Third, it is not clear how to interpret the missing values that originate from a pesticide being regulated in one country but not the other. We cannot replace these missing values with zeros, as is commonly done with missing trade values, because an MRL set to zero is equivalent to banning that pesticide entirely. We choose to fill in these missing values with the maximum MRL (the least restrictive measure) across all importing countries for each product- pesticide pair.                                                              17Only 35 products have pesticide limits regulated by all countries in the sample, among which we find: potatoes, tomatoes, peas, beans, apples, oranges, wheat, maize, sorghum, and ground nuts.  9    For our empirical analysis we use a measure of the relative restrictiveness of pesticide standards, constructed by Ferro et al. (2015), which relies on the restrictiveness of pesticide standards for a product in both the importing/destination country and in the exporting country and is defined as follows: 1 , , , , , , 1 _ , , , , , , , where , , , and , , , are the exporting country c and the destination/importing country d’s MRL for each product k-pesticide a pair in year t, respectively. Thus, , , , , , represents the maximum pesticide standard across all countries while ∈ , , , , , represents the minimum pesticide standard across all countries, ∈ for a given product k-pesticide a pair in year t. The index of relative restrictiveness in Eq. (1) varies between -1 and 1. The index equals 0 when both importing country and exporting country share the same MRLs for a given product, it equals 1 when the importing country has the most restrictive MRLs and the exporting country has the least restrictive MRLs for a product, and it equals -1 when the exporting country has the most restrictive MRLs and the importing country has the least restrictive MRLs for a product.18 If a country does not set an MRL for a given product-pesticide pair - i.e., , , , or , , , are missing - we replace the missing , , , or , , , with MAX , , .19 Larger values of the index indicate that the importing country has a relatively more restrictive standard than the exporting country. An important advantage of our index of relative restrictiveness is that combines into one measure the number of pesticides restricted as well as the intensity with which they are set. Another advantage of our index is that for every product, it includes all pesticides regulated in the world; this contrasts to the limited set of product-pesticide pairs regulated by Codex standards considered by Li and Beghin (2012). Finally, given that some pesticides are more toxic than others, the MRLs for those toxic pesticides are more restrictive in all countries. Therefore it becomes all the more important to normalize the MRLs for each pesticide by a common denominator across all countries and thus to rely on the index of relative restrictiveness in order                                                              18 If MRL=MAX=MIN, for example when only one of the countries regulates a specific product-pesticide pair, the ratio inside the summation takes a value of 1. In absolute terms, a higher value of the index indicates more dissimilar pesticide standards between the importing country and the exporting country.  19 This assumes not setting an MRL for a product-pesticide pair is equivalent to setting the least restrictive MRL across all countries.  10    to compare the restrictiveness of MRL standards across countries. More generally, the use of measures of the relative restrictiveness of pesticide standards is important because it informs on the dissimilarity in the stringency of regulatory requirements across the importing country and the exporting country. As such, it allows us to consider whether the presence of relatively stricter pesticide standards in the importing country imposes additional costs on firms and thus limit their market access.20 That is, our index does capture the presence of a trade barrier given that it accounts for the domestic regulation. It is important to highlight that MRL standards are updated frequently and our estimating strategy takes advantage of this variability over time. Product registrations are withdrawn, new registrations and MRLs are established, and existing MRLs change on a regular basis. Codex MRLs for example, are updated annually every July. New Zealand typically publishes two MRL amendments per year and local officials report that they seek to update MRLs every four months. However, changes in MRLs are not always towards more restrictive standards, but rather very frequently MRLs are increased thus becoming less restrictive. Table 2 shows that on average only 45 percent of MRLs in the dataset are there for the entire panel available for each country. Of those MRLs available for all years, 9 percent became stricter whereas 10 percent became more lenient.21 Analyzing the sources of variation in the index of relative restrictiveness of standards suggests that the main source is variation over time is the intensity of standards. This will be an important element to justify the estimating strategy chosen for the analysis. 3.2 Data on Exporters To measure exporter behavior we use transaction-level customs data for the period 2005- 2012 for 42 developing countries across different regions of the world, obtained from customs agencies (or in a few cases from statistical institutes) collected by the Trade and Integration Unit of the World Bank Research Department, as part of their efforts to expand the Exporter Dynamics Database described in Cebeci et al. (2012). Each country’s raw data-set covers the universe of exporting firms in the agricultural, mining excluding HS Chapter 27 (hydrocarbons                                                              20 Winchester et al. (2012) examine how the dissimilarity in sanitary, phyto-sanitary and conformity requirements (including product requirements such as maximum residue limits for pesticides) across the EU and several of its trading partners affect their bilateral trade but do not but consider the stringency of those requirements, as we do in our study. Disdier et al. (2015) study the impact of standards harmonization promoted in North-South trade agreements and show a negative impact on South-South trade as well as on North-South trade when the harmonization is on regional (rather than international) standards but only harmonization is examined, not the stringency of the standards.  21 The analysis of the sources of variation in the index is described in Appendix A2.  11    such as oil, petroleum, natural gas, coal, etc.), and manufacturing sectors and provides information at the exporter-product-destination-year level for seven variables: country of origin, exporting firm identifier, country of destination, HS 6-digit product, export value, export quantity, and year.22 Additional details on the data are provided in the Appendix. Although for each firm we have information on its exports in all sectors, we define the key outcome variables capturing firm export decisions—firm export value, export quantity, export participation, entry and exit variables—at the exporting country-firm-product-destination- year level focusing exclusively on agricultural and agro-food products, i.e., those belonging to HS Chapters 06-24 (with the exception of HS Chapters 15 and 16), which are the products for which our measures of pesticide standards imposed by importing countries are available. Regarding the intensive margin of trade at the firm level, the outcome variables that we consider are the value and quantity exported by firm i from country c of product k to importing/destination country d in year t which are given by, respectively, ,, , , expressed in current U.S. dollars and ,, , , expressed in kilograms. We also calculate unit values as value divided by quantity expressed in U.S. dollars per kilogram.23 Regarding the extensive margin of trade at the firm level in terms of firms’ decisions to export or to enter or exit a product-destination market, we need to expand (or fill in) the initial dataset described above with only positive exports by adding zeros on some dimensions to be able to define the corresponding outcome variables. If we were to follow the gravity equation literature (particularly studies employing an export participation equation to account for biases in gravity equation estimation) we would expand the initial dataset so as to make it a ‘square’ matrix where every firm in an exporting country would have an observation (a row) for every product-destination-year combination possible. Given the large number of exporting firms in our sample of 42 developing countries, such an expanded data set would be computationally impossible to handle and would be highly cluttered by zeros as most firms tend to export a single product to a single destination. Moreover, our objective in constructing an expanded dataset is to have observations (rows) that make                                                              22Cebeci et al. (2012) show the quality of the data by comparing the 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).  23 The number of observations in specifications that explain quantities or unit values will be smaller than those in the specifications that explain value given that 8 countries in our sample (Botswana, Chile, the Dominican Republic, Guatemala, Mexico, Macedonia, Pakistan and El Salvador) do not include quantity information in their exporter-level customs datasets.  12    economic sense, i.e., that indicate plausible choices for firms without requiring major assumptions. Consider as a first example, an observation from the initial dataset in which firm i starts to export product k to destination d in year t. If in the expanded dataset we add an observation with a 0 export value for firm i product k destination d in year t-1, that implies that in year t-1 we are allowing firm i to choose whether to export product k to destination d and the firm chooses not to do so. This seems like a plausible and not overly restrictive assumption.24 Consider as a second example, firm i exporting products k and l at some point during the sample period in the initial data set. If in the expanded dataset we add observations with 0 export values for firm i for all other possible products (other than k and l) to any possible destination in any year, this implies that in any year we are allowing firm i to choose whether to export any possible agricultural product to all destinations. This seems like an implausible assumption given that other agricultural products may be completely different from what the firm’s capabilities in terms of its technology, type of land, and other inputs allow her to produce (e.g., if a firm produces and exports tropical fruits it is unlikely that the firm can also produce and export wheat or maize which require completely different environmental conditions to grow).25 Our choice is therefore to expand the initial dataset along a dimension that retains computational feasibility, does not require implausible assumptions about the firms’ export choice set, and allows us to exploit an interesting type of variability in the data. We expand the initial dataset so that each firm-product-destination 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- product-destination are not occurring. Setting up the expanded dataset in this way and including a specific (and stringent) type of fixed effects - discussed in Section 4 - allows us to exploit the panel dimension in the firms’ decisions to export, enter or exit a product-destination market as pesticide standards change over time. Using this expanded dataset, we define: - a firm export participation dummy ,, , , equal to 1 in year t if firm i from country c exports a positive value of product k to destination d, and equal to 0 otherwise; - a dummy for firm entry into a product-destination market ,, , , equal to 1 if firm i exports product k to destination d in year t but did not do so in year t-1, and equal to 0 if the firm did                                                              24 The only scenario under which this assumption would be wrong is if the firm only begun to have the capacity (through access to machinery, land, other inputs, etc.) to produce product k in year t not in year t-1 or if the firm did not exist in year t-1.  25 Thus, we would be allowing a firm in a country to have the choice to export products that might not be feasible to grow in that country due to climate and soil conditions.  13    not export product k to destination d in year t-1 and does not start to do so in year t. If the firm continues to export the product-destination market after year t, then the entry dummy becomes missing for years greater than t;26 - a dummy for firm exit from a product-destination market ,, , , equal to 1 if firm i does not export product k to destination d in year t but did so in year t-1 and equal to 0 if the firm exported product k to destination d in year t-1 and continues to do so in year t. Firms that export to a product-destination market in every year are excluded from the entry analysis (since the entry dummy is missing in all of their years) but are included in the analysis of exit and of the intensive margin of trade.27 If a firm has positive exports to a product- destination market only in the first year of the sample (and no exports to that product-destination market in later years of the sample), it is included in the exit analysis but not in the entry analysis because we are unable to determine whether the firm entered the product-destination market in that first year or was already exporting there previously. Alternatively, if a firm has positive exports to a product-destination market only in the last year of the sample (and no exports to that product-destination market in previous years of the sample) it is included in the entry analysis but not in the exit analysis. Table 3 shows for each exporting country the number of firms in our initial dataset. Cambodia and Yemen have the fewest agricultural exporters with less than 50 on average per year, whereas Mexico has the most with an average of 3,313 per year. Table 3 also shows for each exporting country the number of firms in the expanded dataset, which is the same in every year as by construction each firm appearing at least once in the data set has the possibility of exporting in every year in the country’s sample period. Finally, Table 3 shows the total number of observations in the initial dataset that will be used for the intensive margin specifications and the total number of observations in the expanded data set that will be used for the extensive margin specifications. Additional summary statistics on the sample of exporting countries are provided in Appendix C.                                                              26 We follow Koenig (2009), Koenig et al. (2010), and Mayneris and Poncet (2015) and allow for multiple export entries over the sample period for a given firm-product-destination. Multiple entries occur in cases where the firm starts exporting to a product- destination market, then stops, and then re-starts exporting to the same product-destination market.   27 However, in the case of the exit decision, the firms that export to a product-destination market in every year and thus have a 0 in the dependent variable in every year will be effectively dropped from the estimating sample given the specific fixed effects (exporting country-firm-product-destination) included in our specifications.  14    4. Empirical Framework To examine the effects of pesticide standards imposed by importing countries on the export decisions of firms in developing countries exploiting the panel dimension, we consider the following specification: 2 ,, , , ∗ _ , , , ∗ , , , ∗ ,, , , ,, , ,, , , where c is an exporting country, i a firm, k a product, d a destination country, t a year, the dependent variable is either export participation, entry, or exit defined based on the expanded dataset or log export value or log export quantity defined based on the initial dataset. The variable _ , , , is defined in Eq. (1), tariff , , , is the log of 1 plus the bilateral tariff imposed by the destination/importing country d on product k from the exporting country c, and is an independent and identically distributed (i.i.d.) residual.28 The vector ,, , , includes GDP per capita of the destination country and total exports of product k from export country c in year t in all specifications to control for demand and supply effects.29 Applying the logic of the gravity equation to sectoral trade is not entirely straightforward; when looking at sectoral trade flows and particularly agricultural goods, the idea that trade flows between c and d in a certain product k are increasing in the exporter country’s GDP is not necessarily warranted. In the monopolistic-competition model, larger countries produce more varieties of goods and that contributes to increasing their trade. That is, they do not necessarily export more of each good but they export more goods. Additional firm characteristics and other variables included in the regressions will be described in Section 5. A key remark on Eq. (2) concerns the stringent exporting country-firm-product- destination country fixed effects ,, , - for simplicity designated in what follows as firm- product-destination fixed effects given that each firm belongs to a single exporting country - that are included and account for unobserved heterogeneity at that very finely disaggregated level, in addition to the year fixed effects . Accounting for unobserved heterogeneity at such finely                                                              28 Simple average applied tariffs are used for importing country-exporting country pairs for each product and year available in the WITS-TRAINS (Trade Analysis and Information System) database. We interpolate observations to fill in missing years. For cases where applied tariff data is not available for a given importing country-exporting country-product-year cell we replace the missing values with Most Favored Nation (MFN) tariffs of the importing country-product-year or with preferential tariffs of the given importing country-exporting country-product-year for importing country-exporting country pairs which have a preferential tariff agreement. In the export participation regression in Section 5.2 we lose approximately 3% of the observations (13,600) due to the inclusion of tariffs in the regression, as tariff data is missing for 21 agricultural products.   29 GDP per capita data in constant USD is obtained for all countries from the World Development Indicators (WDI) database.  15    disaggregated level may account in particular for the fact that certain pathogens and pests are endemic to certain regions, hence the pesticides used are determined by where the crops grow and thus complying with a given MRL standard can be substantially more costly for a firm producing a similar product using a similar production process and at a similar scale simply due to their specific location. The coefficient of interest is thus identified based on within firm- product-destination changes in export participation, export value, export quantity, entry, or exit as pesticide standards change over time in the destination country relative to the exporting country for that product. The important hypotheses to test in our empirical framework are whether an increase in the stringency of pesticide standards in an importing country relative to an exporting country hampers that exporting country’s firms’ export and entry decisions and encourages their exit decisions. A corollary is that firms in countries with more stringent pesticide standards should have easier market access in destinations with less stringent standards. If relatively more stringent pesticide standards in a destination market increase firms’ variable trade costs to that market, then they would in principle lead to a reduction in firms’ export values (or quantities). But if higher fixed entry costs ensuing from those standards reduced entry into that destination market, incumbent firms’ export values (or quantities) could actually increase. Hence the effect of pesticide standards on the intensive margin of firm exports is theoretically ambiguous. Table 4 displays summary statistics for all firms’ export decisions that will be used as dependent variables as well as for the measure of relative restrictiveness of pesticide standards. 5. Main Results 5.1 Effects of Pesticide Standards on Numbers of Exporters, Entrants, and Exiters As a way to motivate and support the results from estimating Eq. (2) for the extensive margin that will be presented from Section 5.2 onwards, we first analyze the effect of relative product standards on firms’ export decisions aggregated to the level of the exporting country- product-destination country-year level. Specifically, we estimate the equation: 3 , , , ∗ _ , , , ∗ , , , ∗ , , , , , , , , where , , , and are defined as before, and the dependent variable Y is the number of exporters, entrants, and exiters exporting product k from country c to destination country d in 16    year t. These variables are, respectively obtained as the sums at the exporting country-product- destination country-year level of the firm export, entry, or exit dummies defined based on the expanded dataset in Section 3.2. As an alternative to absolute numbers we also allow Y to be exporter entry rates, exit rates, and entrant survival rates. The entry rate is defined as the number of entrants in year t relative to the total number of exporters in the same year. The exit rate in year t is defined as the number of firms that exported in year t-1 but did not export in year t relative to the total number of exporters in year t-1. Similarly, the entrant survival rate in year t is defined as the number of entrants in year t-1 that survive onto year t relative to the number of entrants in year t-1. The vector , , , includes GDP per capita of the destination country and total exports by exporting country-product in each year. Given the fixed effects , , included, the coefficient is identified based on within exporting country-product-destination country changes in numbers of exporters, entrants, or exiters as the corresponding pesticide standards change over time in the destination country relative to the exporting country. The results from estimating Eq. (3) are displayed in Table 5. Poisson regression estimation is used in columns (1)-(3) due to the count nature of the number of exporters, entrants, and exiters while OLS estimation is used in columns (4)-(6) for the rates.30 Column (1) shows that the restrictiveness of importing country standards relative to exporting country standards have a significant negative effect on the number of firms in country c exporting product k to destination d. The number of entrants, the entry and entrant survival rates of exporters also are significantly lower when importing country standards are more restrictive than exporting country standards, in columns (2), (4), and (6). In contrast, significantly higher numbers of firms exit destination markets with more strict standards relative to their own country’s standards in columns (3) and (5). These aggregate results suggest that indeed pesticide standards can be a barrier to enter a market. In the next subsection we directly link firms’ export decisions with the relative restrictiveness of standards in the different actual and potential product-destination markets. 5.2 Effects of Standards on Firms’ Export Decisions                                                              30 The caveat of using OLS is that the predicted rates may lie outside of the [0,1] interval but that is not a limitation to the qualitative predictions from these regressions which are our focus.   17    Tables 6 and 7 present the baseline results from estimating Eq. (2). Inference is based on Huber-White standard errors robust to heteroskedasticity clustered by exporting country- importing country-product-year since the specifications explain firms’ export decisions with our relative restrictiveness index measured at the more aggregate exporting country-product- destination country-year level (Moulton, 1990).31 For the export participation, entry and exit regressions a linear probability model is used in columns (1)-(3) of Table 6 since probit models cannot be estimated with the panel-type of fixed effects considered. We also present robustness results estimated using a conditional fixed effects logit model in columns (4)-(6) of Table 6. For the export value and export quantity regressions OLS estimation is used. The use of linear probability models for the export participation, entry, and exit decisions follows the study by Bernard and Jensen (2004) but has the shortcoming that the predicted probabilities may not be meaningful since they can lie outside of the [0,1] interval and thus the magnitude of the effects of the regressors on those export decisions cannot be assessed. Table 6 shows the estimated effects of pesticide standards on the extensive margin of trade, firms’ export decisions and decisions to enter and/or exit specific product-destination markets. The negative and statistically significant coefficient on Rel_restrictiveness in column (1) suggests firms face a lower probability of exporting when importing country standards are more restrictive than exporting country standards. Similarly, firms’ decisions to start exporting to a new market decrease significantly when the destination’s standards are more stringent than those applied in the domestic market, as seen in column (2). These results are robust when using a conditional logit model in columns (4)-(5). There are several reasons why more restrictive standards in the importing country decrease the likelihood that a firm from an exporting country with laxer standards enters that market. First, the asymmetry of information that new exporters to a market suffer from, relative to domestic producers or incumbent exporters, constitutes an additional cost to potential exporters that need to collect the information on the regulations imposed across foreign markets. Second, pesticide standards vary greatly from country to country and producers need to adapt production methods to meet the standards imposed by each destination market. Stricter standards are likely to be harder to meet and thus to require a greater investment by producers in order to                                                              31 We also considered robust standard errors clustered by exporting firm and our results were robust. Those results are available from the authors upon request.  18    comply. As our results show, stricter standards in importing countries relative to exporting countries result in fewer producers attempting to enter those importing countries’ markets. Columns (3) and (6) of Table 6 show how pesticide standards affect firms’ decisions to stop exporting a product to a destination country. Our results show that stricter standards in the importing country relative to the exporting country are not significantly associated with the likelihood that an exporter will exit that market. Our conjecture was that as the stringency of standards increases firms would have to stop exporting to the market as many would not be able to comply with that change in standards. Some potential rationales for our insignificant findings are that incumbent exporters to a market may receive the necessary information regarding changes in regulations in a timely manner so as to be able to adjust their production process in order to comply with the new regulation without having to stop exporting or being denied entry at the border for not meeting the adequate standards. It is also possible that because more restrictive standards deter entry, incumbent exporters enjoy lower competition and are able to pass-through the cost of regulatory compliance onto consumers and avoid exit. We will, however, show in Section 6 that this insignificant effect of standards on exit is not robust. Table 7 shows the results for the intensive margin of trade, which should be interpreted as the effect of pesticide standards on exports, conditional on there being positive exports. The estimates show that more restrictive standards imposed by the importing country relative to the exporting country lead to significantly lower firm export values and quantities. The effect on firm export unit values is not significant. More stringent standards imposed by the importing country can result in higher quality products being imported, which would be reflected in higher unit prices, but in the case of our exporting countries it appears that the negative effect of the restrictiveness of standards on export quantities dominates any change in unit prices, thus the overall effect of the restrictiveness of standards on export values is negative.32 5.3 Further Effects of Standards on Export Entry for Existing Exporters In this section we explore the mechanisms that may lie behind the effects of product standards on firms’ decisions to enter a new product-destination market, focusing on firm                                                              32 Notice that the sample used for the quantity and the export unit price regressions in columns (2) and (3) of Table 7 is smaller than that used for the export value regression in column (1) due to the lack of quantity information for 8 countries. Unreported results from estimating the export value regression using the same sample as in columns (2) and (3) show a negative and significant coefficient on Rel_restrictiveness. The regression results available from the authors upon request.   19    characteristics and network effects. Trade models with heterogeneous firms suggest that trade policies or regulations may affect firms’ export performance differentially depending on firms’ characteristics. The impact of product standards on firms’ export decisions may depend on firm size, which is likely to be associated with firm productivity, and hence with a firm’s ability to overcome additional costs to export. In that setup relatively more restrictive importing country product standards could have a greater detrimental impact on small exporters. Additionally, standards could affect firms entering a new product-destination market differently than firms already exporting product k and beginning to export it to a new destination d, or firms already exporting to destination d and beginning to export product k there. Network effects may also play a role, i.e., the presence of other firms from the same exporting country in a given destination may alleviate the negative impact of product standards on export entry. A first network effect may occur as firms from a given exporting country receive information - including on product standards - about possible product-destination markets through firms already established in those markets.33 A second network effect may occur when foreign buyers or distribution networks in an importing country attempt to expand imports to levels that previous exporters alone are unable to meet, and hence new firms are approached to export to that destination. A third possible spillover effect of incumbent exporters on potential entrants is through the availability of inputs in the domestic market, such as replacement pesticides for banned ones or the necessary human capital and/or know-how to implement new production methodologies to meet more stringent foreign standards. Table 8 shows the results from estimating several variants of Eq. (2) where the measure Rel_restrictiveness enters by itself but also interacted with different measures of firm size and firm size also enters by itself. It is important to highlight that even though we are analyzing the effect of standards on firms’ entry decisions, we focus here on a particular type of entry in year t for existing exporters (i.e., firms which were already exporting in year t-1): firms’ decisions to enter into new products, new destinations, and/or new product-destinations (depending on the measure of firm size). The reason for doing this is that our proxy for firm size is firm total export value, for which we choose to consider the one-year lag in order to allow past performance to affect current export decisions and to mitigate endogeneity concerns. Thus, the entirely new                                                              33Cadot et al. (2013) show that survival in export markets for African firms rises with the number of firms from the same country exporting the same product to the same destination, with one of the key channels for the effects being the sharing of information among exporters already established in a market.  20    exporters in year t, i.e., those firms that did not export any product to any destination in year t-1 are excluded from the analysis here (though they were included in the estimating sample for entry regressions in Table 6).34 Column (1) shows results for a regression where size is measured by the (logarithm of) the firm’s total agricultural exports, Agro_exportsc,i,t-1, to all of its destinations (including those for which we do not have MRL data and thus are not part of our sample of importing countries).35 Hence, this regression estimates the effect of relative standards on entry into new products, new destinations, or new product-destinations. The coefficient on Rel_restrictiveness is negative and significant as in Table 6, indicating that stricter pesticide standards in the importing country relative to the exporting country reduce the likelihood of a firm entering that product- destination market, even after controlling for firm size. Also, all else equal larger firms are more likely to enter into a new product-destination market than smaller ones. Finally, our main coefficient of interest on the interaction between Rel_restrictiveness and firm size (Agro_exports) is positive and significant indicating that larger exporters are less negatively impacted by the relative stringency of pesticide standards. This is consistent with the empirical evidence that larger exporters tend to be more productive and thus are able to absorb higher costs related to compliance with foreign product standards. Hence, the size of the firm appears to be a direct mechanism through which the relative restrictiveness of standards affects a firm’s decision to export to a new market. Column (2) displays results for a regression where size is measured by the (logarithm of the) firm’s exports of agricultural products to destination country d, Agro_exports_destc,i,d,t-1.36 This regression estimates the effect of relative standards on firms’ decisions to introduce new products into existing destinations. The coefficient on Rel_restrictiveness remains negative and significant, indicating that stricter pesticide standards in the importing country relative to the exporting country reduce the likelihood of a firm introducing a new product into an existing destination, even after controlling for firm size. The coefficient on firm size is positive and significant, indicating a firm is more likely to introduce a new product into a destination country                                                              34 Consequently, the number of observations in all entry regressions in Table 8 is substantially smaller than that in the entry regression in column (2) of Table 6.  35 Firm size is obtained as the sum of total agricultural exports in HS chapters 06-24 with the exception of chapters 15 and 16. While the variable entering the regression is the one-year lag of firm size, unreported regressions including the current firm size obtain qualitatively similar results. The results available from the authors upon request.   36 All firms which did not already export agricultural products other than k to destination d in year t-1 are dropped from the estimating sample for this regression.  21    in which it is already a large exporter. The interaction term between Rel_restrictiveness and firm size (Agro_exports_dest) has an insignificant coefficient. This lack of significance is actually intuitive. The advantage that larger firms have over smaller firms in meeting restrictive standards can dissipate if the firm is already exporting to the destination. If the advantage of large exporters to meet standards in foreign markets is due to the fact that they can more easily overcome information asymmetries regarding those standards, then already serving a given destination country can help reduce those information asymmetries, even for small firms. Column (3) shows results where size is measured by the (logarithm of the) firm’s total exports of product k, Exports_productc,i,k,t-1.37 This regression allows us to analyze the effects of relative standards on firms’ decisions to enter into a new destination with an existing export product. The effect of Rel_restrictiveness remains negative and significant, indicating that stricter pesticide standards in the importing country relative to the exporting country reduce the likelihood of a firm diversifying to new destinations, even after controlling for firm size. Larger firms are also more likely to enter to new destinations. The positive and significant coefficient on the interaction term between Rel_restrictiveness and firm size (Exports_product) indicates that the negative effect of relative standards on the likelihood of entering new destinations is not as strong for firms that are already large exporters of a product. The regression in column (4) of Table 8 is estimated for a specific measure of entry that identifies the existing exporters that enter into a new product-destination market in year t. The results show that stricter pesticide standards in the importing country relative to the exporting country reduce firms’ likelihood of entering into new product-destination markets. The positive coefficient on firm size suggests that larger firms are more likely to enter into new product- destination markets than smaller firms. Moreover, larger firms are less affected in their market entry decisions by the negative effects of more stringent relative restrictiveness of product standards. The regressions in Table 9 explore further channels through which firms might be able to reduce information asymmetries about foreign markets such as network effects, while also controlling for firm size. Column (1) reproduces the results shown in column (1) of Table 8 while columns (2) and (3) show the role of the network effects proxied either by the (logarithm of the) number of firms from the same country as firm i that export (any) agricultural products to                                                              37 All firms which did not already export product k in year t-1 are dropped from the estimating sample for this regression.  22    each destination in year t-1, Number_agro_firmsc,d,t-1 in column (2) or by the (logarithm of the) number of firms from the same country exporting the same product k to the same destination d as firm i in year t-1, Number_prod_firmsc,k,d,t-1 in column (3). The estimates show that firms are more likely to enter into a new destination where more firms are already selling agricultural products or where more firms are already selling the exact same product. The coefficient on Rel_restrictiveness is negative and significant in columns (2) and (3). The coefficient on the interaction term between Rel_restrictiveness and Number_agro_firms in column (2) is insignificant suggesting that the number of exporters of agricultural products to a given destination does not mediate the effect of restrictive standards of the importing country on entry into that destination. The interaction term between Rel_restrictiveness and Number_prod_firms in column (3) is positive and significant at the 10% level suggesting that firms can to some extent overcome the negative effect of the relative restrictiveness of destination country standards when more exporters from their home country export the same product to that destination. These results provide some evidence of positive network effects for firms’ decisions to start exporting to a new destination in helping them overcome the regulatory impediments of restrictive standards in a new destination, but only when firms export the same product. These network effects might be related with information asymmetries considering that information on foreign standards might flow from one firm to another either through labor mobility or through producer associations. Also, the likelihood that a firm can find the necessary inputs and techniques to meet foreign standards will likely increase with the number of home country firms already exporting the same product to a given destination. 5.4 Further Effects of Standards on Export Exit In this section we explore whether product standards impact firms’ decisions to stop exporting to destinations where they previously sold agricultural products differentially depending on firm size, the product’s importance in firm total exports, or the destination’s importance in firm total exports. The rationale behind the consideration of the product’s importance is that a firm which derives most of its export revenue from one product is more likely to incur in the costs necessary to comply with changes in standards in order to sustain its main source of revenue. In contrast, a firm which sells multiple products is more likely to stop 23    exporting a product that is not important for its overall export portfolio if it is more costly to comply with changes in standards due to the small scale of production. A similar rationale applies to the consideration of the destination country’s importance. The exit regressions focus by definition on the decision to stop exporting to a product-destination market for existing exporters. Table 10 shows the results from estimating several variants of Eq. (2) where the measure Rel_restrictiveness enters by itself but also interacted with firm size (Agro_exportsc,i,t-1) as well as with the share in the firm’s total agricultural exports of exports of the product, Product_sharec,i,k,t-1, or exports of the destination, Destination_sharec,i,d,t-1. Firm size and the shares of the product or destination in firm total exports also enter by themselves. Column (1) explores the role of firm size for the decision to exit a product-destination market. In contrast to what was shown in the baseline estimates in Table 6, once firm size is controlled for, the coefficient on Rel_restrictiveness is positive and significant, indicating that firms are more likely to stop exporting a product to a destination with relatively more restrictive standards than their domestic market. The coefficient on firm size shows that larger firms are less likely to exit their export markets. The interaction term between Rel_restrictiveness and firm size has a negative and significant coefficient indicating that larger firms are less likely to exit a product-destination market due to the relative restrictiveness of that importing country’s pesticide standards. Column (2) explores the importance of the product while column (3) explores the importance of the destination, in the firm’s total agricultural exports. In both columns the coefficient on Rel_restrictiveness is positive and significant, again in contrast to the finding in Table 6. Also, our estimates show that, all else equal, the more important is a given product or destination to a firm, the less likely is the firm to stop exporting that product or to that destination, respectively. The interaction terms between Rel_restrictiveness and either Product_share or Destination_share have negative coefficients, suggesting a less severe exit- inducing effect of more stringent standards for products or destinations that account for a larger share of firms’ export portfolios, but the effects are not significant. Column (4) presents the results from a specification which controls for firm size, product share, destination share, as well as the interaction between each of these variables and the Rel_restrictiveness measure. The coefficient on Rel_restrictiveness is positive and significant while that on its interaction with firm size is negative and significant as in column (1). The 24    effects of the interaction between Rel_restrictiveness and either the importance of the product or the destination market for firms’ total agricultural exports are negative but insignificant. Therefore, firm size is the channel by which the relative restrictiveness of pesticide standards in the importing country affects the firms’ decision to stop exporting to a product-destination market. 6. Robustness Results In this section we describe a number of robustness checks that we performed to our results. We re-estimated the export participation, entry, and exit regressions including interactions with firm size for three variants of the sample. The first set of regressions whose results are shown in columns (1)-(3) of Table 11 restricts the sample of importing countries and exporting countries to be only those that draft specific MRL regulations, that is, the sample excludes the countries that use solely Codex standards as their own regulations. The effect of the importing country’s relative restrictiveness of standards on firms’ export participation and entry into new destinations remains negative and significant while the effect on firms’ exit is positive and significant. The coefficient on the interaction between Rel_restrictiveness and firm size is significant in both the entry and exit specifications suggesting that the effect of pesticide standards on firms’ export decisions depends on firm size. Smaller firms are more negatively affected by the relative stringency of importing country standards than larger firms. The second set of regressions whose results are shown in columns (4)-(6) of Table 11 restricts the sample in terms of the years covered, keeping only years 2008-2012 and thus addressing the fact that in 2006 and 2007 there is missing standards data for a few destinations including the EU, as seen in Table 2. Again, the results show that the relative restrictiveness of standards of the importing country negatively impacts a firm’s decision to export and firm size determines how much the standards affect the firm’s decision. The third set of regressions whose results are shown in columns (7)-(9) of Table 11 restricts the sample of products to those belonging to HS chapters 7 and 8 (edible vegetable and edible fruits). The match on MRL standards obtained from Homologa database was better for these two HS chapters. Again all results are qualitatively maintained. We conduct one final robustness check in which we include additional sets of stringent fixed effects in order to control for unobservables that might not be fully controlled for in our 25    baseline estimation.38 In addition to including in all specifications the panel-type firm-product- destination fixed effects, we include the set of importing country-year, exporting country-year, and product-year fixed effects in the first variation in columns (1), (4), (7), and (10) of Table 12, importing country-product-year fixed effects in columns (2), (5), (8), and (11) of Table 12, and exporting country-product-year fixed effects in columns (3), (6), (9), and (12) of Table 12. These various types of fixed effects are more flexible than the specific control variables included in our previous specifications (destination country GDP per capita or exporting country-product total exports). For example exporting country-year and destination country-year fixed effects may account for the differential effects of macroeconomic shocks in the exporting and the destination country and are particularly relevant for us to account for given that our sample period encompasses the recent Global Financial Crisis (GFC). Also, exporting country-product- year fixed effects account for supply shocks such as weather shocks affecting a product’s crop whereas destination country-product-year fixed effects account for demand shocks such as an increased preference for certain types of fruits. The estimates in Table 12 show that the conclusions remain the same as in the baseline specification: the relative restrictiveness of importing country standards affects negatively firms’ export participation and entry into new markets and affects positively firms’ exit from a given market. The results also confirm that larger firms can more easily overcome the negative effects of standards. 7. Conclusion This paper examines the importance of standards’ regulations in influencing the ability of firms in developing countries to exploit export opportunities for agricultural and food products. Our evidence shows that the relative restrictiveness of importing country pesticide standards restricts market access for those firms. Importantly, our findings suggest a significant effect of restrictive importing country standards in depressing both the firm extensive and intensive margins of exports. Our results are consistent with recent trade models which predict that only the most productive firms are able to overcome the fixed costs of exporting. Obtaining information on foreign standards and adjusting production processes to comply with those                                                              38 In unreported results we also re-estimated the export participation, entry, and exit decisions including instead of GDP per capita of the destination country and total exports by export country-product as in Eq. (2) the more traditional gravity regressors of GDP of the destination country and GDP of the export country as an alternative way to control for demand and supply effects. The effects of the importing country’s relative restrictiveness of standards were qualitatively maintained.   26    standards increase the fixed costs to reach foreign markets. This helps explain the stronger negative effect of the relative restrictiveness of an importing country’s standards on entry and exit decisions for smaller firms. The international community has attempted to overcome the trade-distortive effects of standards through the WTO SPS Agreement and other trade tools. This has included decades of discussion about the benefits of harmonization of standards, use of international consensus standards through international bodies such as Codex Alimentarius, and potential cost of domestic standards that deviate from international norms. Limited progress has been made, however, in these various steps to mitigate the negative affect of discriminatory or duplicative national standards. Moreover, a number of developing countries lack access to compliance resources, including scientific and technical expertise, information and finance to exploit the opportunities offered in various trade agreements. Looking ahead, it is important that new trade talks consider and address the channels through which NTMs affect trade. SPS standards, including MRLs, continue to be developed by national governments and this heterogeneity is likely to remain in place since full global harmonization of such standards is impossible. Furthermore, the development of new and deeper trade agreements in which NTMs are included—which is likely to be the case for the TTIP—will greatly impact third-country firms in developing countries that may be unable to meet the new agreed-upon standards. For a developing country seeking to support its agricultural exporters entering and surviving in markets with more stringent SPS standards, the best strategy is for the government to help exporters meet foreign standards by creating the necessary testing facilities and customs clearance procedures, and by facilitating the imports of the necessary inputs to meet foreign standards. Future research on the effects of NTMs on trade should expand the analysis to the manufacturing sector using measures that account not only for the number of standards but also for their stringency, as we have done in this paper with standards for agricultural products. In addition, while our study focuses on the impact of “de jure” regulations, the implementation of these regulations at ports and border crossing points may or may not be fully enforced by the importing country’s authorities. Investigating those effects is left for future work. References Anders, S. and J. Caswell (2009), ‘Standards as Barriers versus Standards as Catalyst: Assessing the Impact of HACCP Implementation on U.S. Seafood Imports’, American Journal of Agricultural Economics, 91(2), 310–321. 27    Bernard, A. and B. Jensen (2004), ‘Why Some Firms Export?’, Review of Economics and Statistics, 86(2), 561-569. Bernard A., S. Redding and P. Schott (2011), ‘Multi-Product Firms and Trade Liberalization’, Quarterly Journal of Economics 126(3), 1271-1318. Cadot and Malouche (2012), Non-Tariff Measures- A Fresh look at Trade Policy’s New Frontier. CEPR and The World Bank. Cato, J., W. S. Otwell and A. Coze (2004), ‘Nicaragua’s Shrimp Subsector: Developing a Production Capacity and Export Market during Rapidly Changing Worldwide Safety and Quality Regulations’, Agriculture and Rural Development Discussion Papers (47844), The World Bank. Cebeci, T., A. Fernandes, C. Freund and M. Pierola (2012), ‘Exporter Dynamics Database’, Policy Research Working Paper No. 6229, The World Bank. Chaney, T. (2008), ‘Distorted Gravity: The Intensive and Extensive Margins of International Trade’, American Economic Review, 98(4), 1707-1721. Chen, X., J. Wilson and T. Otsuki (2010), ‘Standards and Export Decisions: Firm-Level Evidence from Developing Countries’, Journal of International Trade and Economic Development, 17(4), 501-523. Chen, C. and C. Findlay (2008), ‘Measuring the Effect of Food Safety Standards on China’s Agricultural Exports’, Review of World Economics, 144(1), 83-106. Czubala, W., B. Shepherd and J. S. Wilson (2009), ‘Help or Hindrance? The Impact of Harmonized Standards on African Exports’, Policy Research Working Paper No. 4400, The World Bank. Disdier, A-C., L. Fontagné and O. Cadot (2015), ‘North-South Standards Harmonization and International Trade’, World Bank Economic Review, 29(2), 327-352. Disdier, A-C., L. Fontagné and M. Mimouni (2008), ‘The Impact of Regulations on Agricultural Trade: Evidence from the SPS and TBT Agreements’, American Journal of Agricultural Economics, 90(2), 336–350. Ferro, E., J. S. Wilson and T. Otsuki (2015), ‘The Effect of Product Standards on Agricultural Exports’, Food Policy 50, 68-79. Fontagné, L., G. Orefice, R. Piermartini and N. Rocha (2015), ‘Product Standards and Margins of Trade: Firm Level Evidence’,Journal of International Economics forthcoming. Henson S., A-M. Brouder and W. Mitullah (2000), ‘Food Safety Requirements and Food Exports from Developing Countries: The Case of Fish Exports from Kenya to the European Union’, American Journal of Agricultural Economics, 82(5), 1159-1169. Cadot, O., Iacovone, L., Pierola, M., and F. Rauch (2013), “Success and Failure of African Exporters”; Journal of Development Economics 101, 284–296. Koenig, P., F. Mayneris and S. Poncet (2010), ‘Local Export Spillovers in France’, European Economic Review, 54(4), 622-641. Koenig, P., (2009), ‘Agglomeration and the Export Decisions of French Firms’, Journal of Urban Economics, 66(3), 186–195. Li, Y. and J. Beghin (2012), ‘Protectionism Indices for Non-Tariff Measures: An Application to Maximum Residue Levels’, Iowa State Working Paper No.12013. 28    Maskus, K., T. Otsuki and J. S. Wilson (2005), ‘The Cost of Compliance with Product Standards for Firms in Developing Countries: An Econometric Study’, Policy Research Working Paper No. 3590, The World Bank. Maertens, M. and J. F. M. Swinnen (2009), ‘Trade, Standards and Poverty: Evidence from Senegal’, World Development, 37(1), 161–178. Mayneris, F. and S. Poncet (2015), ‘Chinese Firms’ Entry to Export Markets: The Role of Foreign Export Spillovers’, World Bank Economic Review, 29(1), 150-179. Melitz, M. (2003), ‘The Impact of Trade on Intra-Industry Reallocations and Aggregate Industry Productivity’, Econometrica, 71(6), 1695-1725. Moulton, B. (1990), ‘An Illustration of a Pitfall in Estimating the Effects of Aggregate Variables on Micro Units’, Review of Economics and Statistics, 72(2), 334-38. Otsuki, T., J. S. Wilson and M. Sewadeh (2001a), ‘Saving Two in a Billion: Quantifying the Trade Effect of European Food Safety Standards on African Exports’, Food Policy, (26), 495–514. Otsuki, T., J. S. Wilson and M. Sewadeh (2001b), ‘What Price Precaution? European Harmonization of Aflatoxin Regulations and African Groundnut Exports’, European Review of Agriculutural Economics, 28 (2), 263-283. Shepherd, B. and N. Wilson (2013), ‘Product Standards and Developing Country Agricultural Exports: The Case of the European Union’, Food Policy, 42, 1–10. Tran, N., N. Wilson and S. Anders (2011), ‘Standard Harmonization as Chasing Zero (Tolerance Limits): The Impact of Veterinary Drug Residue Standards on Crustacean Imports in the EU, Japan and North America’, American Journal of Agricultural Economics, 94(2), 496–502. Wilson, J. S., T. Otsuki and B. Majumdsar (2003), ‘Balancing Food Safety and Risk: Do Drug Residue Limits Affect International Trade in Beef?’, Journal of International Trade and Economic Development, 12(4), 377–402. Winchester, N., M. L. Rau, C. Goetz, B. Larue, T. Otsuki, K. Shutes, C. Wieck, H. L. Burnquist, M. J. Pinto de Souza and R. N. de Faria (2012), ‘The Impact of Regulatory Heterogeneity on Agri-Food Trade’, The World Economy, 35(8), 973-993. Xiong, B. and J. Beghin (2010), ‘Does European Aflatoxin Regulation Hurt Groundnut Exporters from Africa?’, European Review of Agricultural Economics, 39(4), 589-609. 29    Table 1. Number of Products Regulated by Pesticide Standards by Importing Country Number of Specific Products Regulated  Number of  Products Regulated or With Default MRL 2006 2007 2008 2009 2010 2011 2012 2006 2007 2008 2009 2010 2011 2012 Argentina 102 110 110 110 115 109 110 214 219 232 240 242 241 241 ASEAN  ‐ ‐ 102 102 102 102 102 ‐ ‐ 147 148 149 149 148 Australia 139 154 155 152 152 152 152 139 154 155 152 152 152 152 Brazil 59 66 66 72 73 73 73 147 152 141 145 149 150 149 Canada 118 142 157 160 161 164 165 214 219 232 240 242 241 241 Chile 146 149 136 136 138 85 85 146 149 136 137 142 143 142 China 87 87 87 43 43 44 44 87 87 87 43 43 44 44 CODEX 146 149 136 137 140 141 140 146 149 136 137 140 141 140 Colombia ‐ 136 136 136 136 136 137 ‐ 149 136 137 140 141 141 Dom. Republic 146 149 136 137 140 141 140 183 192 191 191 192 193 192 Egypt ‐ ‐ ‐ 136 136 136 137 ‐ ‐ ‐ 240 242 241 241 European Union ‐ ‐ 125 135 136 136 136 ‐ ‐ 232 240 242 241 241 Honduras 146 149 136 137 140 141 140 183 192 191 191 192 193 192 India ‐ 134 149 105 105 105 105 ‐ 149 150 151 151 152 151 Israel 93 87 87 87 87 87 87 149 152 139 140 143 144 143 Japan 129 117 115 119 116 116 106 214 219 232 240 242 241 241 Republic of  Korea 92 92 100 98 100 100 100 92 92 100 98 100 100 100 Malaysia 42 86 86 84 84 84 88 214 219 232 240 242 241 241 Mexico 66 66 72 67 67 67 67 167 184 182 180 181 182 182 Morocco 146 149 136 137 140 141 140 146 149 171 175 177 177 176 New Zealand 142 109 99 86 86 86 86 214 219 232 240 242 241 241 Panama 146 149 136 137 140 141 140 183 192 203 205 206 207 206 Russian Federation 32 95 113 112 113 111 111 32 95 113 112 113 111 111 Singapore ‐ ‐ 125 125 125 125 125 ‐ ‐ 150 151 151 152 151 South Africa 89 99 99 99 99 99 99 214 219 232 240 242 241 241 Switzerland 125 127 133 133 145 141 141 125 127 147 150 146 142 142 Taiwan, China 68 68 75 77 77 97 97 68 68 75 77 77 97 97 Thailand 102 102 102 102 102 102 124 154 157 149 150 153 154 157 Turkey 103 103 103 98 137 134 134 103 103 134 137 138 136 136 Ukraine ‐ ‐ ‐ 114 114 114 114 ‐ ‐ ‐ 114 114 114 114 United States 166 183 180 180 181 182 182 166 183 180 180 181 182 182 Total 214 219 232 240 242 241 241 214 219 232 240 242 241 241 Note: ‘-‘ indicates that the country does not have data for that particular year. ASEAN is the Association of Southeast Asian Nations. 30    Table 2. Changes in Pesticide Standards over Time by Importing Country Number of  Share  of  Of Observations with Complete  Panel   Observations  Observations  (MRLs available  in every year) (at country‐HS  with Complete   Share  with  product‐ Panel  (MRLs  Share  with  Share  with  More   pesticide ‐year  available  in  No Change   Stricter  Lenient  level) every year) in MRLs MRLs MRLs Argentina 90,084 49% 68% 27% 5% ASEAN  5,193 75% 85% 11% 4% Australia 11,508 46% 90% 4% 6% Brazil 20,865 23% 63% 8% 29% Canada 90,084 49% 97% 1% 2% Chile 21,540 16% 42% 11% 46% China 2,714 32% 88% 3% 8% CODEX 19,674 16% 42% 12% 45% Colombia 19,668 20% 82% 3% 23% Dom. Republic 25,875 28% 47% 14% 39% Egypt 87,314 93% 96% 3% 2% European Union 88,333 87% 92% 4% 4% Honduras 25,875 28% 47% 14% 39% India 20,970 23% 54% 11% 40% Israel 21,804 26% 71% 7% 22% Japan 90,084 49% 86% 9% 5% Republic of Korea 13,240 40% 87% 9% 4% Malaysia 90,084 49% 67% 27% 6% Mexico 10,585 49% 79% 9% 11% Morocco 64,918 27% 71% 10% 19% New Zealand 90,084 49% 96% 3% 1% Panama 67,566 30% 65% 17% 18% Russian Federation 10,490 1% 74% 18% 8% Singapore 6,945 82% 93% 5% 2% South Africa 90,084 49% 63% 4% 32% Switzerland 67,619 20% 66% 18% 16% Taiwan, China 5,862 67% 93% 6% 1% Thailand 19,924 20% 58% 9% 33% Turkey 65,225 18% 87% 7% 6% Ukraine 3,385 100% 99% 0% 0% United States 10,180 46% 75% 12% 13% Total 1,257,776 45% 81% 9% 10% Note: ASEAN is the Association of Southeast Asian Nations. 31    Table 3. Number of Firms and Observations by Exporting Country Number of Exporting Firms by Exporting Country‐Year  Number of Observations by Exporting Country Time ‐Series                         Initial  Dataset Initial  Dataset  Expanded Dataset 2005 2006 2007 2008 2009 2010 2011 2012 Across All Years Per Year Across All  Years Albania 43 81 77 74 77 90 99 85 1,164 644 5,152 Burkina Faso ‐ ‐ 129 136 109 128 131 160 1,629 1,210 7,260 Bangladesh 243 383 321 408 343 355 429 ‐ 6,584 3,680 25,760 Bulgaria 516 779 ‐ ‐ ‐ ‐ ‐ ‐ 9,049 6,565 13,130 Bolivia ‐ 241 252 245 256 244 230 213 5,515 2,352 16,464 Botswana 70 90 99 109 118 113 ‐ ‐ 1,881 1,149 6,894 Chile 790 963 981 1,018 1,037 1,092 1,155 1,029 54,127 20,024 160,192 Cote  d'Ivoire ‐ ‐ ‐ ‐ 310 300 332 374 5,315 3,351 13,404 Cameroon ‐ ‐ 111 138 137 ‐ ‐ ‐ 973 696 2,088 Colombia ‐ ‐ 817 756 708 ‐ ‐ ‐ 7,713 4,648 13,944 Costa Rica 540 690 698 707 705 747 729 627 20,838 8,671 69,368 Dom. Republic ‐ ‐ 610 469 580 480 547 475 21,891 12,633 75,798 Ecuador ‐ 426 458 488 542 ‐ ‐ ‐ 6,932 3,831 15,324 Ethiopia ‐ ‐ ‐ 542 692 799 706 652 9,437 5,964 29,820 Georgia 95 176 117 132 130 150 151 163 2,581 1,716 13,728 Guatemala 617 738 797 782 767 809 ‐ ‐ 20,342 8,780 52,680 Croatia ‐ ‐ 243 236 235 238 254 273 4,241 1,954 11,724 Iran ‐ 1,525 1,521 1,472 1,233 1,277 ‐ ‐ 18,726 13,021 65,105 Jordan 79 110 122 125 117 133 ‐ ‐ 1,580 1,341 8,046 Kenya ‐ 781 600 649 683 ‐ ‐ ‐ 6,974 4,545 18,180 Kyrgyzstan ‐ 85 93 96 134 131 117 99 2,057 1,190 8,330 Cambodia ‐ ‐ 44 35 45 ‐ ‐ ‐ 199 157 471 Lebanon ‐ ‐ ‐ 402 381 388 398 401 16,001 9,658 48,290 Morocco 452 637 642 651 665 696 700 614 17,773 8,095 64,760 Madagascar ‐ ‐ 187 215 224 246 357 225 4,396 2,769 16,614 Mexico 1,522 4,236 3,508 3,604 3,693 ‐ ‐ ‐ 48,279 28,681 143,405 Macedonia 204 328 349 319 322 355 ‐ ‐ 6,388 3,509 21,054 Mauritius 59 85 96 96 97 96 92 87 3,634 1,773 14,184 Malawi ‐ 81 87 73 86 82 108 111 1,616 628 3,943 Nicaragua 150 218 217 248 245 253 240 ‐ 5,419 2,676 18,732 Pakistan 805 1,110 1,133 1,239 1,502 1,541 ‐ ‐ 21,914 12,559 75,354 Peru 643 980 1,009 1,053 1,122 ‐ ‐ ‐ 18,275 9,927 49,635 Paraguay ‐ ‐ 181 193 204 196 193 186 3,349 1,549 9,294 Romania 346 500 312 394 464 659 775 10,998 6,443 45,101 Senegal ‐ ‐ ‐ 117 121 132 142 153 1,512 1,028 5,140 El  Salvador 157 197 193 198 220 ‐ ‐ ‐ 3,136 1,420 7,100 Tanzania 172 205 206 232 263 275 292 4,681 2,742 19,194 Uganda ‐ ‐ 184 164 172 232 ‐ ‐ 2,287 1,433 5,732 Uruguay 146 174 182 169 190 201 218 187 5,268 2,392 19,136 Yemen ‐ ‐ ‐ 49 35 28 19 19 240 167 835 South Africa 1,019 1,416 1,411 1,457 1,453 1,414 1,390 1,839 107,562 49,039 392,312 Zambia 56 83 129 104 135 180 212 ‐ 1,706 1,126 7,882 Total 8,724 17,318 18,116 19,594 20,552 14,060 10,016 7,972 494,182 249,171 1,600,559 Note: - indicates that the country does not have data for that particular year. 32    Table 4. Summary Statistics on Key Dependent and Independent Variables Number of  Mean Std. Dev. Minimum Maximum Observations  Export Participation Dummyc,i,k,d,t 796,636 0.32 0.47 0 1 Entry Dummyc,i,k,d,t 535,056 0.22 0.41 0 1 Exit Dummyc,i,k,d,t 240,074 0.45 0.50 0 1 Log Export Value c,i,k,d,t  256,882 9.25 3.40 ‐9.06 20.91 Log Export Quantityc,i,k,d,t 159,914 8.57 3.79 ‐6.91 20.24 Log Export Unit Price c,i,k,d,t 159,914 0.51 1.25 ‐16.12 13.17 Rel_restrictivenessc,k,d,t 796,636 ‐0.06 0.52 ‐1 1 Note: log export value or quantity are defined based on the initial data set whereas export participation, entry, or exit are defined based on the expanded data-set. Export value is measured in USD, export quantity in kilograms, and export unit value in USD per kilogram. Table 5. Effects of Standards on Numbers of Exporters, Entrants, and Exiters Dependent Variable: Number of  Number of  Number of  Exporter entry  Exporter exit  Entrant suvival   exportersc,k,d,t entrantsc,k,d,t exitersc,k,d,t rate c,k,d,t rate c,k,d,t rate c,k,d,t (1) (2) (3) (4) (5) (6) Rel_restrictivenessc,k,d,t ‐0.247 ‐0.445 0.486 ‐0.044 0.050 ‐0.060 [0.040]*** [0.072]*** [0.090]*** [0.019]** [0.020]** [0.035]* Log 1+Tariffc,k,d,t ‐0.112 ‐0.161 ‐0.093 ‐0.013 ‐0.019 0.005 [0.014]*** [0.024]*** [0.024]*** [0.006]** [0.006]*** [0.010] Log GDP per capitad,t 0.976 2.174 1.417 0.001 0.359 0.148 [0.171]*** [0.244]*** [0.263]*** [0.046] [0.052]*** [0.075]** Log Total  Exportsc,k,t 0.199 0.224 ‐0.007 ‐0.007 ‐0.028 0.033 [0.008]*** [0.011]*** [0.009] [0.003]** [0.003]*** [0.004]*** Exporting Country‐ Product‐Destination  Yes Yes Yes Yes Yes Yes Country Fixed Effects Year Fixed Effects  Yes Yes Yes Yes Yes Yes Estimation Method Poisson Poisson Poisson OLS OLS OLS Observations 82,793 78,871 69,214 48,365 46,302 31,484 R‐squared 0.573 0.589 0.566 Notes: Robust standard errors in brackets. OLS standard errors are clustered at the exporting country-importing country-product level. ***, **, and * indicate significance at 1%, 5%, and 10% confidence levels, respectively. 33    Table 6. Effects of Standards on Firm Export Participation, Entry, and Exit Dependent Variable:   Export  Export        Export         Export  Export        Export          Participation  Entry     Exit            Participation  Entry     Exit            Dummyc,i,k,d,t  Dummyc,i,k,d,t  Dummyc,i,k,d,t   Dummyc,i,k,d,t  Dummyc,i,k,d,t  Dummyc,i,k,d,t   (1) (2) (3) (4) (5) (6) Rel_restrictivenessc,k,d,t ‐0.074 ‐0.032 0.009 ‐0.425 ‐0.237 ‐0.098 [0.011]*** [0.016]** [0.031] [0.036]*** [0.049]*** [0.182] Log 1+Tariffc,k,d,t ‐0.030 ‐0.025 0.015 ‐0.192 ‐0.195 0.053 [0.005]*** [0.007]*** [0.008]* [0.011]*** [0.013]*** [0.046] Log GDP per capitad,t 0.260 0.008 ‐0.549 1.621 0.494 ‐5.863 [0.036]*** [0.047] [0.095]*** [0.091]*** [0.105]*** [0.465]*** Log Total  Exportsc,k,t 0.047 0.045 ‐0.071 0.276 0.256 ‐0.352 [0.002]*** [0.002]*** [0.005]*** [0.005]*** [0.006]*** [0.021]*** Exporting Firm‐Importing  Yes  Yes  Yes  Yes  Yes  Yes  Country‐Product Fixed Effects Year Fixed Effects  Yes  Yes  Yes  Yes  Yes  Yes  Linear  Linear  Linear  Conditional   Conditional   Conditional   Estimation method Probability  Probability  Probability  Logit Logit Logit Model   Model   Model   Observations 796,636 535,056 240,074 634,930 396,153 93,606 R‐squared 0.351 0.242 0.703 Pseudo R‐squared 0.013 0.031 0.302 Notes: Robust standard errors in brackets, clustered at the exporting country-importing country-product-year level. ***, **, and * indicate significance at 1%, 5%, and 10% confidence levels, respectively. Dependent variables are defined as follows: export participation equals one if there is positive trade between countries and zero otherwise; entry equals one if firm i exports product k to destination d in year t but did not do so in year t-1 and zero if the firm did not export product k to destination d in year t-1 and does not start to do so in year t. Similarly, exit equals one if firm i does not export product k to destination d in year t but did so in year t-1 and zero if the firm exported product k to destination d in year t-1 and continues to do so in year t. 34    Table 7. Effects of Standards on Firm Export Value, Quantity, and Unit Price Dependent Variable: Log Export  Log Export  Log Unit  Value c,i,k,d,t  Quantityc,i,k,d,t  Price c,i,k,d,t  (1) (2) (3) Rel_restrictivenessc,k,d,t ‐ 0.173 ‐0.179 0.020 [0.057]*** [0.065]*** [0.032] Log 1+Tariff c,k,d,t ‐ 0.045 ‐0.071 ‐0.003 [0.016]*** [0.023]*** [0.011] Log GDP per capitad,t 1.340 0.852 0.182 [0.164]*** [0.189]*** [0.078]** Log Total  Exportsc,k,t 0.321 0.232 0.045 [0.015]*** [0.016]*** [0.012]*** Exporting Firm‐Importing  Yes Yes Yes Country‐ Product Fixed Effects Year Fixed Effects  Yes Yes Yes Estimation method OLS OLS OLS Observations 256,882 159,914 159,914 R‐squared 0.948 0.959 0.922 Notes: Robust standard errors in brackets, clustered at the exporting country-importing country-product-year level. ***, **, and * indicate significance at 1%, 5%, and 10% confidence levels, respectively. 35    Table 8. Effects of Standards on Firm Export Entry: Role of Firm Size Dependent Variable: Export Entry  Export Entry  Export Entry  Export Entry  Dummyc,i,k,d,t  Dummyc,i,k,d,t  Dummyc,i,k,d,t  Dummyc,i,k,d,t  (1) (2) (3) (4) Rel_restrictivenessc,k,d,t ‐0.190 ‐ 0.109 ‐0.253 ‐0.180 [0.025]*** [0.034]*** [0.034]*** [0.072]** Log Agro_exportsc,i,t‐1 0.051 0.030 [0.001]*** [0.003]*** Rel_restrictivenessc,k,d,t X Log Agro_exportsc,i,t‐1 0.010 0.011 [0.002]*** [0.005]** Log Agro_exports_destc,i,d,t‐1 0.040 [0.002]*** Rel_restrictivenessc,k,d,t X Log Agro_exports_destc,i,d,t‐1 0.000 [0.002] Log Exports_productc,i,k,t‐1 0.044 [0.002]*** Rel_restrictivenessc,k,d,t X Log Exports_productc,i,k,t‐1 0.021 [0.003]*** Log 1+Tariffc,k,d,t ‐0.020 ‐ 0.012 ‐0.017 ‐0.050 [0.004]*** [0.006]** [0.009]* [0.012]*** Log GDP per capitad,t 0.030 ‐ 0.336 0.296 0.087 [0.042] [0.051]*** [0.061]*** [0.114] Log Total  Exportsc,k,t 0.041 0.038 0.047 0.043 [0.002]*** [0.002]*** [0.004]*** [0.007]*** Exporting Firm‐Importing Country‐Product Fixed Effects Yes Yes Yes Yes Year Fixed Effects  Yes Yes Yes Yes Linear  Linear  Linear  Linear  Estimation method Probability  Probability  Probability  Probability  Model   Model   Model   Model   Observations 297,409 168,151 132,626 40,436 R‐squared 0.379 0.465 0.406 0.635 Notes: Robust standard errors in brackets, clustered at the exporting country-importing country-product-year level. ***, **, and * indicate significance at 1%, 5%, and 10% confidence levels, respectively. The dependent variable is defined as follows: entry equals one if firm i exports product k to destination d in year t but did not do so in year t-1 and zero if the firm did not export product k to destination d in year t-1 and does not start to do so in year t. 36    Table 9. Effects of Standards on Firm Export Entry: Role of Firm Size and Network Effects Dependent Variable: Export Entry  Export Entry  Export Entry  Dummyc,i,k,d,t  Dummyc,i,k,d,t  Dummyc,i,k,d,t  (1) (2) (3) Rel_restrictivenessc,k,d,t ‐0.190 ‐0.152 ‐0.186 [0.025]*** [0.050]*** [0.026]*** Log Agro_exportsc,i,t‐1 0.051 0.050 0.049 [0.001]*** [0.001]*** [0.001]*** Rel_restrictivenessc,k,d,t X Log Agro_exportsc,i,t‐1 0.010 0.009 0.009 [0.002]*** [0.002]*** [0.002]*** Number_agro_firmsc,d,t‐1 0.110 [0.010]*** Rel_restrictivenessc,k,d,t X Number_agro_firmsc,d,t‐1 ‐0.007 [0.010] Number_product_firmsc,d,  k,  t‐1 0.079 [0.005]*** Rel_restrictivenessc,k,d,t X Number_product_firmsc,d,k,t‐1 0.007* [0.006] Log 1+Tariffc,k,d,t ‐0.020 ‐0.016 ‐0.015 [0.004]*** [0.004]*** [0.004]*** Log GDP per capitad,t 0.030 ‐0.100 ‐0.009 [0.042] [0.038]*** [0.038] Log Total  Exportsc,k,t 0.041 0.041 0.038 [0.002]*** [0.002]*** [0.002]*** Exporting Firm‐Importing Country‐Product Fixed Effects Yes Yes Yes Year Fixed Effects  Yes Yes Yes Linear  Linear  Linear  Estimation method Probability  Probability  Probability  Model   Model   Model   Observations 297,409 297,409 297,409 R‐squared 0.379 0.381 0.382 Notes: Robust standard errors in brackets, clustered at the exporting country-importing country-product-year level. ***, **, and * indicate significance at 1%, 5%, and 10% confidence levels, respectively. The dependent variable is defined as follows: entry equals one if firm i exports product k to destination d in year t but did not do so in year t-1 and zero if the firm did not export product k to destination d in year t-1 and does not start to do so in year t. 37    Table 10. Effects of Standards on Firm Export Exit: Role of Firm Size and Network Effects Dependent Variable: Export Exit  Export Exit  Export Exit  Export Exit  Dummyc,i,k,d,t   Dummyc,i,k,d,t   Dummyc,i,k,d,t   Dummyc,i,k,d,t   (1) (2) (3) (4) Rel_restrictivenessc,k,d,t 0.413 0.450 0.470 0.505 [0.128]*** [0.131]*** [0.137]*** [0.140]*** Log Agro_exportsc,i,t‐1 ‐0.093 ‐0.095 ‐0.095 ‐0.096 [0.003]*** [0.003]*** [0.003]*** [0.003]*** Rel_restrictivenessc,k,d,t X Log Agro_exportsc,i,t‐1 ‐0.033 ‐0.033 ‐0.036 ‐0.036 [0.009]*** [0.009]*** [0.009]*** [0.009]*** Product_share c,i,k,t‐1 ‐0.084 ‐0.076 [0.037]** [0.037]** Rel_restrictivenessc,k,d,t X Product_share c,i,k,t‐1 ‐0.070 ‐0.069 [0.046] [0.046] Destination_share c,i,d,t‐1 ‐0.084 ‐0.076 [0.017]*** [0.017]*** Rel_restrictivenessc,k,d,t X Destination_share c,i,d,t‐1 ‐0.032 ‐0.031 [0.046] [0.046] Log 1+Tariffc,k,d,t 0.008 0.008 0.007 0.007 [0.010] [0.010] [0.010] [0.010] Log GDP per capitad,t ‐0.603 ‐0.600 ‐0.593 ‐0.592 [0.112]*** [0.114]*** [0.112]*** [0.113]*** Log Total  Exportsc,k,t ‐0.073 ‐0.072 ‐0.073 ‐0.073 [0.006]*** [0.006]*** [0.006]*** [0.006]*** Exporting Firm‐Importing Country‐Product Fixed  Yes Yes Yes Yes Effects Year Fixed Effects  Yes Yes Yes Yes Linear  Linear  Linear  Linear  Estimation method Probability  Probability  Probability  Probability  Model   Model   Model   Model   Observations 240,074 240,074 240,074 240,074 R‐squared 0.709 0.709 0.709 0.709 Notes: Robust standard errors in brackets, clustered at the exporting country-importing country-product-year level. ***, **, and * indicate significance at 1%, 5%, and 10% confidence levels, respectively. The dependent variable is defined as follows: exit equals one if firm i does not export product k to destination d in year t but did so in year t-1 and zero if the firm exported product k to destination d in year t-1 and continues to do so in year t. 38    Table 11. Results from Robustness Checks – Restricted Samples Dependent Variable: Export  Export        Export         Export  Export        Export          Export  Export        Export          Participation  Entry     Exit            Participation  Entry     Exit            Participation  Entry     Exit            Dummyc,i,k,d,t  Dummyc,i,k,d,t  Dummyc,i,k,d,t   Dummyc,i,k,d,t  Dummyc,i,k,d,t  Dummyc,i,k,d,t   Dummyc,i,k,d,t  Dummyc,i,k,d,t  Dummyc,i,k,d,t   Destination‐Restricted Sample Year‐Restricted Sample Product‐Restricted Sample (1) (2) (3) (4) (5) (6) (7) (8) (9) Rel_restrictivenessc,k,d,t ‐0.116 ‐0.247 0.302 ‐0.069 ‐0.104 ‐0.148 ‐0.117 ‐0.109 0.603 [0.015]*** [0.039]*** [0.160]* [0.031]** [0.044]** [0.203] [0.018]*** [0.040]*** [0.179]*** Log Agro_exportsc,i,  t‐1 0.048 ‐0.089 0.047 ‐0.098 0.053 ‐0.096 [0.002]*** [0.003]*** [0.001]*** [0.004]*** [0.001]*** [0.003]*** Rel_restrictivenessc,k,d,t X           Log Agro_exportsc,i,t‐1 0.007 ‐0.029 0.005 0.016 0.003 ‐0.044 [0.002]*** [0.011]** [0.002]** [0.013] [0.002] [0.013]*** Log 1+Tariffc,k,d,t ‐0.027 ‐0.048 ‐0.023 ‐0.029 ‐0.001 0.004 ‐0.029 ‐0.018 ‐0.005 [0.008]*** [0.014]*** [0.021] [0.007]*** [0.007] [0.015] [0.008]*** [0.007]*** [0.012] Log GDP per capitad,t 0.225 0.192 ‐0.361 0.407 0.199 ‐0.052 0.387 0.185 ‐0.762 [0.067]*** [0.077]** [0.185]* [0.054]*** [0.056]*** [0.140] [0.056]*** [0.062]*** [0.141]*** Log Total  Exportsc,k,t 0.042 0.038 ‐0.077 0.046 0.040 ‐0.068 0.054 0.048 ‐0.082 [0.002]*** [0.003]*** [0.009]*** [0.002]*** [0.002]*** [0.008]*** [0.003]*** [0.003]*** [0.009]*** Exporting Firm‐Importing  Yes Yes Yes Yes Yes Yes Yes Yes Yes Country‐Product  Fixed Effects Year Fixed Effects  Yes Yes Yes Yes Yes Yes Yes Yes Yes Linear  Linear  Linear  Linear  Linear  Linear  Linear  Linear  Linear  Estimation method Probability  Probability  Probability  Probability  Probability  Probability  Probability  Probability  Probability  Model   Model   Model   Model   Model   Model   Model   Model   Model   Observations 441,796 157,212 118,455 608,241 236,042 143,881 478,185 167,137 128,989 R‐squared 0.359 0.395 0.697 0.439 0.471 0.745 0.347 0.386 0.705 Notes: Robust standard errors in brackets, clustered at the exporting country-importing country-product-year level. ***, **, and * indicate significance at 1%, 5%, and 10% confidence levels, respectively. Dependent variables are defined as follows: export participation equals one if there is positive trade between countries and zero otherwise; entry equals one if firm i exports product k to destination d in year t but did not do so in year t-1 and zero if the firm did not export product k to destination d in year t-1 and does not start to do so in year t. Similarly, exit equals one if firm i does not export product k to destination d in year t but did so in year t-1 and zero if the firm exported product k to destination d in year t-1 and continues to do so in year t. 39    Table 12. Results from Robustness Checks – Alternative Fixed Effects Dependent Variable: Log Export  Log Export  Log Export  Export  Export  Export  Export        Export        Export        Export         Export         Export         Value c,i,k,d,t  Valuec,i,k,d,t  Value c,i,k,d,t  Participation  Participation  Participation  Entry     Entry     Entry     Exit            Exit            Exit            Dummyc,i,k,d,t  Dummyc,i,k,d,t  Dummyc,i,k,d,t  Dummyc,i,k,d,t  Dummyc,i,k,d,t  Dummyc,i,k,d,t  Dummyc,i,k,d,t   Dummyc,i,k,d,t   Dummyc,i,k,d,t   (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Rel_restrictivenessc,k,d,t ‐0.155 ‐ 0.158 ‐ 0.382 ‐0.030 ‐ 0.068 ‐ 0.178 ‐0.042 ‐ 0.272 ‐0.177 0.255 0.200 0.328 [0.071]** [0.125] [0.080]*** [0.008]*** [0.013]*** [0.009]*** [0.029] [0.043]*** [0.031]*** [0.098]*** [0.115]* [0.105]*** Log Agro_exportsc,i,  t‐1 0.039 0.041 0.039 ‐0.041 ‐ 0.037 ‐0.045 [0.001]*** [0.001]*** [0.001]*** [0.003]*** [0.003]*** [0.003]*** Rel_restrictivenessc,k,d,t X  Log Agro_exportsc,i,t‐1 0.003 0.008 0.001 ‐0.021 ‐ 0.017 ‐0.023 [0.002]* [0.002]*** [0.002] [0.007]*** [0.008]** [0.007]*** Log 1+Tariff c,k,d,t ‐0.027 ‐ 0.003 ‐ 0.089 ‐0.009 0.003 ‐ 0.054 ‐0.017 ‐ 0.002 ‐0.038 ‐0.004 ‐ 0.004 0.027 [0.019] [0.039] [0.017]*** [0.002]*** [0.005] [0.002]*** [0.005]*** [0.013] [0.004]*** [0.008] [0.020] [0.007]*** Exporting Firm‐ Importing  Country‐ Product  Fixed  Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Effects Exporting Country‐Year,  Importing Country Year,  Yes No No Yes No No Yes No No Yes No No and Product‐ Year Fixed  Importing Country‐ Product‐ Year Fixed  No Yes No No Yes No No Yes No No Yes No Effects  Exporting Country‐ Product‐ Year Fixed  No  No Yes No  No Yes No  No Yes No  No Yes Effects  Linear  Linear  Linear  Linear  Linear  Linear  Linear  Linear  Linear  Estimation method OLS OLS OLS Probability  Probability  Probability  Probability  Probability  Probability  Probability  Probability  Probability  Model   Model   Model   Model   Model   Model   Model   Model   Model   Observations 261,661 261,661 261,661 827,878 827,878 827,878 282,957 282,957 282,957 216,938 216,938 216,938 R‐ squared 0.949 0.956 0.951 0.365 0.408 0.375 0.417 0.514 0.445 0.759 0.805 0.763 Notes: Robust standard errors in brackets, clustered at the exporting country-importing country-product-year level. ***, **, and * indicate significance at 1%, 5%, and 10% confidence levels, respectively. Dependent variables are defined as follows: export participation equals one if there is positive trade between countries and zero otherwise; entry equals one if firm i exports product k to destination d in year t but did not do so in year t-1 and zero if the firm did not export product k to destination d in year t-1 and does not start to do so in year t. Similarly, exit equals one if firm i does not export product k to destination d in year t but did so in year t-1 and zero if the firm exported product k to destination d in year t-1 and continues to do so in year t. 40