TRADE FACILITATION AND GLOBAL VALUE CHAINS IN A POST-PANDEMIC WORLD

The objective of this paper is twofold. First, it examines the effects of the implementation of trade facilitation (TF) measures on international trade flows and on participation in global value chains (GVCs). Second, it provides policy recommendations for developing Asia and the Pacific derived from model estimations. The main focus is to disentangle the effectiveness of a wide range of TF actions taken at a country level and see whether these have paved the way toward more sustainable trade flows. The data used are from the UN Global Surveys on Digital and Sustainable Trade Facilitation, covering 144 countries worldwide for the years 2015, 2017, 2019, and 2021. The empirical estimations obtained from a gravity model of trade indicate that TF measures related to transparency, institutions, and formalities are of utmost importance, whereas sustainable TF actions are still in their infancy, and more data are needed for a proper evaluation of their effectiveness. Moreover, the implementation of TF is related to reductions in the time and cost to export and import as well as to improvements in logistic performance. There are several policy recommendations and implications for developing Asia and the Pacific concerning TF, trade, and GVCs: first, transparency policies will increase trade in manufactured goods more than proportionally, especially in low-income countries in Asia (such as Afghanistan, Bangladesh, Bhutan, Cambodia, Lao PDR, Myanmar, Nepal, and Timor-Leste); second, improving the quality of TF-related institutions will foster participation in global value chains; and finally, for small islands in the Pacific (Kiribati, Solomon Islands, Tuvalu, and Vanuatu), public investments directed at improving their logistics performance would be of special relevance.


INTRODUCTION
In the 2000s, the world economy experienced a sharp increase in international trade flows, as well as an increasing diversification of traded goods and services.Substantial improvements in information technology and sustained economic growth have been important factors contributing to this increase in trade (Xing, Gentile, and Dollar 2021).In particular, part of this new trade has been in intermediate products with the development of global value chains (GVCs).However, this trend was interrupted by the outbreak of the economic crisis of 2008-2009 and more so by that of the COVID-19 pandemic in 2020.One consequence of the latter crisis was the disruptions in GVCs driven by the lockdowns and the subsequent wish to relocate production to nearby countries.On the one hand, these disruptions were related to the pandemic outbreaks happening in different countries simultaneously and the lack of workers in important industries related to GVCs, such as logistics and transport.On the other hand, these phenomena could have accelerated the ongoing transformation in the logistics branch with reinforced support for paperless trade, electronic documents, and the automation of customs procedures.In relation to this, the ratification of the Trade Facilitation Agreement in 20171 should have started to pave the road that countries have to follow to adopt the required measures for implementation.
The main contribution of this paper is in quantifying the advantages derived from improving trade facilitation measures that help reduce trade barriers among countries, including time delays and administrative burdens for the products and services exchanged.This quantification should help to disentangle the effect of different components of trade facilitation in increasing gross trade and trade in value added after COVID-19, particularly in Asia and the Pacific countries.The main question to be answered is whether a number of trade facilitation measures implemented by countries have contributed to increasing bilateral and aggregated trade, as well as to the participation of countries in GVCs, and if so, to what extent.
There is scant empirical research covering the periods before and after the Trade Facilitation Agreement that includes value-added exports in the analysis.For instance, while some authors focused on the effects of transport infrastructure on trade (Limao and Venables 2001;Márquez-Ramos et al. 2011) and on trade facilitation issues (Wilson, Mann, and Otsuki 2003;Engman 2005;Persson 2007;Martínez-Zarzoso and Márquez-Ramos 2008;Hendy and Zachi 2021;Shepherd 2022;Kareem, Martínez-Zarzoso, and Bruemer 2022), only a few of them focused on value-added exports in developing countries (Xu, Sun, and Jiang 2022;Zhang and Martínez-Zarzoso 2022), and none of them covered recent years and aspects related to trade finance, support for small and medium-sized enterprises (SMEs), agriculture, and women.We will be able to include these factors in the analysis by making use of a new data set based on the UN Global Surveys on Digital and Sustainable Trade Facilitation, covering 144 countries worldwide for the years 2015, 2017, 2019, and 2021.This paper aims to narrow this gap in the literature by investigating the relationship between trade facilitation and exports (gross exports and GVC participation) for a global sample of countries and a recent period of time (2015)(2016)(2017)(2018)(2019)(2020)(2021) that covers the COVID-19 pandemic.The empirical methodology consists in applying a gravity model of bilateral trade estimated using panel data methods that control for unobserved heterogeneity using multidimensional fixed effects and panel data models with countryfixed effects for aggregated trade flows and GVC participation and position.
The main results show that most of the trade facilitation indicators considered, namely those related to transparency, paperless trade, institutions, and formalities, have a direct influence on trade flows and some of them on participation in GVCs.However, there are insufficient data to assess the importance of sustainable trade facilitation, given that the data collection for such indicators started in 2020 and there is no information for many countries.Moreover, transparency policies will increase trade in manufactured goods more than proportionally, especially in low-income countries in the Asia and the Pacific region.Second, improving the quality of TF-related institutions will foster participation in global value chains in the region.Finally, public investments directed towards improving their logistics performance would be of special relevance for small islands.
The rest of the paper is organized into five sections.Section 2 presents a review of the literature on trade facilitation.Section 3 describes the data and variables used and some stylized facts.Section 4 describes the empirical strategy and the econometric estimation techniques used.Section 5 presents and discusses the main results from estimating a gravity model of bilateral trade and panel data models for aggregated trade and GVC indicators.Finally, Section 6 concludes and presents some policy implications.

LITERATURE REVIEW
The international trade literature has focused widely on trade facilitation issues since the early 2000s.Seminal contributions by Wilson, Mann, andOtsuki (2003, 2005) used a wide definition of trade facilitation and claimed that trade in the Asia and the Pacific region could increase by 21% if low performers improved their scores halfway to the average.Instead, Engman (2005) used the World Trade Organization (WTO) definition of trade facilitation, simplifying and harmonizing international trade procedures involving activities at the border, while also finding positive impacts on trade.A number of authors2 investigated the effects of specific measures, including administrative barriers (Hummels and Schaur 2013;Djankov, Freund, and Pham 2010;Hendy and Zaki 2021), information technology (Márquez-Ramos et al. 2007;Rodriguez-Crespo and Martínez-Zarzoso 2021), port efficiency (Limao and Venables 2001;Martínez-Zarzoso and Hofmann 2007;Wilmsmeier, Martínez-Zarzoso, and Fiess 2011), maritime networks (Márquez-Ramos et al. 2011), and the quality of institutions (Gylfason, Martínez-Zarzoso, and Wijkman 2015;Martínez-Zarzoso and Márquez-Ramos 2019).The main takeaway from the existing literature is that advances in trade facilitation actions foster international trade to some extent, with the key issue being quantifying the effects to ascertain what measures are more effective.More recently, some authors have focused on the effects of TF on GVC participation.In this respect, Kumar and Shepherd (2019) find that the full implementation of the TF agreement will increase trade by about 3.5% with respect to 2015 and could lead to changes in the composition of trade, promoting trade in intermediates and hence the development of value chain trade mostly in middle-income economies.Moreover, Shepherd (2022) investigates the effect of changes in trade facilitation performance on changes in GVC trade.He uses the TF indicators computed by the OECD for the period 2015-2019 and finds that in some sectors, the estimated elasticity of TF on trade is higher for intermediates than for total trade, with the quantitative differences being small.In addition, several works have jointly estimated the effect of trade facilitation variables and policy trade barriers (Márquez-Ramos, Martínez-Zarzoso, and Suárez-Burguet 2012 Hendy and Zachi 2021, among others), showing that the former are in general more important than the latter for trade.Finally, some recent papers have specifically focused on Asia or Asian subregions.More specifically, Central Asia was the focus of Kim, Mariano, and Abesamis (2022) and Cheong and Turakulov (2022), whereas Ismail (2021) investigated digital trade facilitation in selected Asian countries and Ramasamy and Yeung (2019) analyzed the impact of trade facilitation in relation to the People's Republic of China's (PRC) One Belt, One Road initiative.Also, Halaszovich and Kinra (2020) present some insights into trade facilitation in Asia, indicating that the elements of national transportation systems positively influence both trade and foreign direct investment.
With regard to the empirical methodologies, two main modeling strategies have been used.First, a number of authors relied on estimating a gravity model of trade, which includes trade facilitation factors in the specification as proxies for trade easiness (Wilson, Mann, andOtsuki 2003, 2005;Djankov, Freund, and Pham 2010;Nordas, Pinali, and Grosso 2006;Soloaga, Wilson, and Mejía 2006;Persson 2007;Martínez-Zarzoso and Márquez-Ramos 2008;Kumar and Shepherd 2019;Kim, Mariano, and Abesamis 2022;Shepherd 2022).Second, several works (Decreux and Fontagne 2006;Dennis 2006;Cheong and Turakulov 2022, among others) used computable general equilibrium models to estimate the effect of trade facilitation indices on trade flows.Overall, independently of the approach used, the results of the studies show positive and statistically significant effects derived from improved trade facilitation on international trade.
The present paper departs from existing literature in two respects.First, it focuses on both bilateral and aggregated trade and on GVCs; and second, it analyzes the effect of newly collected TF measures in the most recent years and after the TF agreement with a special focus on developing countries in the Asia and the Pacific region.It also provides policy recommendations for developing Asia and the Pacific derived from the model estimations, disentangling the effectiveness of a wide range of TF actions, taken at a country level.

DATA, VARIABLES, AND STYLIZED FACTS
This section first presents the data, variables, and sources (3.1) and next the stylized facts (3.2).

Data and Variables
Export and import data at the bilateral level are from UNCTAD, and data on aggregated exports and imports of goods and services, as well as GDP and GDP per capita at constant prices, are from the World Development Indicators data set.Other gravity variables, namely geographical distance, and whether countries share a common language, common border, and have or have had a colonial relationship, are extracted from the Centre d'Études Prospectives et d'Informations Internationales (CEPII).The Doing Business data set from the World Bank is the source for the number of documents needed to trade, time to trade, and cost to trade across countries, with data available using the new methodology from 2014 to 2019.Data for regional trade agreements (RTAs) are from De Sousa (2012), updated by Martínez-Zarzoso and Chelala (2021), using information from the WTO.Trade facilitation data are from the United Nations Global Surveys on Digital and Sustainable Trade Facilitation. 3hey cover 144 countries worldwide for the years 2015, 2017, 2019, and 2021.The 2021 survey includes 58 questions4 that are listed in the Appendix (Table A.1).The implementation stage and rate of implementation for each measure are provided for selected groups of trade facilitation factors.The grouped TF measures are transparency, formalities, institutions, paperless, cross-border, transit, TF for SMEs, agriculture, and women.Each of them varies from zero to 100, indicating the rate of implementation (zero = no implementation),5 and is composed of a number of subcomponents as indicated in Table A.1.For instance, the transparency measure includes questions Q2-Q5 and Q9.The first four indicate the advance publication of trade-related regulations on the Internet and stakeholders' consultation, and the fifth whether there is an independent appeal mechanism.The original measures are coded with values from 0 to 3, indicating whether the measure: has been not implemented, is in the pilot stage of implementation, has been partially implemented, or has been fully implemented, respectively.
The survey was led and coordinated by ESCAP and jointly conducted in 2021 by five United Nations Regional Commissions: ECA, ECE, ECLAC, ESCAP, and ESCWA.Data for 2021 are duplicated for countries that participated in the 2019 Global Survey but did not answer in 2021 (Antigua and Barbuda, Belize, Brazil, El Salvador, Guyana, Saint Vincent and the Grenadines, Trinidad and Tobago, Tanzania, and Tunisia).
Proxies for GVCs are constructed using information from the UNCTAD-Eora GVC database that covers the period from 1990 to 2018, which is the last year available, to decompose gross exports (see Koopman, Wang, and Wei 2014;Wang et al. 2017;Borin and Mancini 2019).The gross exports' components are value-added exports (VA), foreign value added (FVA), domestic value added (DVA), and domestic value added in exports (DVX), available for 189 countries. 6 country's GVC participation (GVCP) is measured as a share of its gross exports (UNCTAD 2013).For country and year, the GVC participation index is given by: where FVAit denotes foreign value added in country i at time t.DVXit denotes domestic value added in exports in country i at time t.
A second variable constructed is the GVC position index, which indicates the relative "upstreamness" of a country in a GVC (Koopman, Wang, and Wei 2014).Examples of upstream activities are branding, design, and research and development, with all of them being capital-intensive pre-production activities that require high-skilled labor.
Otherwise, downstream activities are associated with post-production services of high value added, such as sales and marketing.The GVC position of a country in a given year is given by: where FVAit denotes foreign value added in country i at time t.DVXit denotes domestic value added in exports in country i at time t.

Stylized Facts
In this section, we present some figures that show the implementation stage of TF measures in 2021.Figures 1 and 2 show the rate of implementation in 2021 of the TF aggregated indicators in the Asia and the Pacific region by subregions and in the world economy compared with the Asia and the Pacific region, respectively.Figure 1 indicates that East Asia is the best performer in terms of transparency, transit, and formalities, with TF measures reaching more than 80% of the full implementation target (scale between 0 and 1), but the same subregion is the worst performer in terms of sustainable TF measures (related to women, sustainable agriculture, and SMEs).
Central and West Asia are also doing well in regard to transparency, but not so concerning cross-border TF, as they have the lowest degree of implementation (below 30%).As regards Southeast Asia, sustainable TF is also poorly implemented, whereas transparency and formality TF measures show a better achievement.The Pacific does best in transparency TF and worst in cross-border TF measures, whereas South Asia shows 80% implementation in transparency, and around 70% in formality, institutions, and paperless TF-adopted measures.3), paperless trade (Figure 4), and formalities (Figure 5). Figure 3 indicates that the best score in transparency items is for countries in East Asia, whereas the worst refers to countries in the Pacific subregion.This means that while the transparency TF measures have been fully implemented in the former, they are still in the planning stage or have been partially implemented in the latter region.In regard to paperless TF measures, Figure 4 shows that there is room for improvement in most subregions concerning the implementation of electronic single windows and several electronic application processes, which are still not fully implemented in any of the considered subregions.Similarly, Figure 5 shows that formalities for TF are, at best, partially implemented, indicating that risk management in South Asia is either in the planning stage or partially implemented, as is the case for post-clearance audits in the Pacific.Note: Average scores are calculated only using economies where measure implementation information is available (i.e., not implemented = 0, planning stage = 1, partially implemented = 2, or fully implemented = 3).

EMPIRICAL STRATEGY
To evaluate the effects of TF measures across countries, in light of the TF agreement, the main strategy consists in: first, estimating a gravity model of bilateral trade for the years 2016 to 2021 for a global sample of countries and for exports of primary and manufactured products separately; second, estimating correlations between total exports of goods and services and GVC proxies for a global sample of countries and for specific geographical areas, with a special focus on the Asia and the Pacific region; third, considering the doing business time to exports and cost to export and import as dependent variables.

The Gravity Model of Trade
The gravity model of trade has been extensively used to estimate the factors that explain bilateral trade flows among countries (Feenstra 2004).In the last four decades, being a structural model with firm theoretical foundations, it has been considered the workhorse for international trade analysis, as documented by Eaton and Kortum (2002), Anderson and Van Wincoop (2003), Allen, Arkolakis, and Takahashi (2014), Head and Mayer (2014), and Anderson, Larch, and Yotov (2018), among others.Therefore, it is suitable for evaluating the effects of trade facilitation factors.
The international trade theories in relation to the gravity model were reformulated and modernized by Anderson and Van Wincoop (2003).The model's underlying assumptions are that the elasticity of substitution between goods is constant and products are differentiated by country of origin.Moreover, bilateral trade costs are assumed to be symmetric, so prices differ among countries.According to the gravity model, bilateral exports between two countries are directly proportional to the product of their economic mass and inversely proportional to the costs of trade between them.
The multiplicative form of the model is given by: where Xijt is the bilateral exports from country i to country j in year t, and Yit (Yjt) and Y W are the GDP of the exporting (importing) country and the world in year t, respectively.tijt is the trade costs between the pair of trading countries in year t, and Pit and Pjt are price indices that reflect the multilateral resistance terms (MRTs).σ is the elasticity of substitution between goods.
The model specification in its log-linear form is given by: The presence of trade costs and MRTs in Equation ( 4) implies that some estimation issues must be considered.For instance, the trade cost function tijt, is generally assumed to be a function of several trade barriers.These include the geographical distance between countries, the lack of a common border, a common colonial past and common language (all time-invariant), and a number of policy variables, including membership in multilateral agreements such as: regional trade agreements (RTAs), World Trade Organization (WTO), and trade facilitation variables (all time-varying).The trade cost function is given by: Substituting Equation (5) into Equation ( 4) and extending the model with year dummy variables and an error term gives the next model specification: where Xijkt denotes exports of shipped from country i to country j in year t, lnDij denotes the natural logarithm of the distance between country i and country j, and TFit (TFij) denotes trade facilitation measures taken by country i (j) at time t.Contigij takes the value of 1 for a pair of countries sharing a border, and 0 otherwise.Comlangij and Comcolij take the value of 1 when a pair of countries share an official language or have ever had a colonial relationship, respectively, and 0 otherwise; RTAijt takes the value of 1 when the trading countries are members of a regional trade agreement, and 0 otherwise; WTOijt takes the value of 1 if country i or country j is a WTO member and 2 if both are members.t denotes a set of year dummies that proxy for business cycle and other time-variant common factors (globalization) that affect all trade flows in the same manner.
The MRTs are modeled using time-invariant country-specific dummies (i, j), given the short time span of our sample and the year-time variation of our TF variables.In the final specification, the time-invariant gravity variables that account for trade cost factors are substituted by country-pair fixed effects   to control for all bilateral unobserved characteristics.The model is specified as: In this regard, we follow Baier and Bergstrand (2007) and Head and Mayer (2014), who suggested the use of pair fixed effects and time dummy variables to control for bilateral unobserved heterogeneity and common time trends, respectively.For completeness, we also include estimates of the traditional gravity model ( 6) that include economic and bilateral variables and with common time effects in the Appendix (Table A.3 for manufactures and Table A.4 for primary products).
According to Head and Mayer (2014), fixed effects that vary by exporter-time (it) and importer-time (jt) could be included as a proxy for MRTs.In this case, variables such as GDP and TF cannot be identified directly. 7As a way to identify the effect of variables that vary by country and over time, such as TF, a two-stage approach is used (also following Martínez-Zarzoso and Chelala 2020).In the first stage, the country-time fixed effects are estimated from the following gravitational model: Exporter-time (  ) and importer-time (  ) fixed effects represent trade barriers that are country-specific and vary over time, that is, third-party countries' barriers to trade that affect the costs of trade.
The exporter-time and importer-time fixed effects, extracted from Model (8), are used as dependent variables in the second stage: To account for factors such as institutions, infrastructure, and cultural factors that vary slowly, the estimation includes unobservable country effects,   .Yit indicates the exporter's GDP.TF takes the value of 1 when the exporting country has applied a given TF improvement in period t.X denotes additional control variables that have country-time variation.
In addition to the log-linearized models proposed, and based on the ongoing development of new techniques for estimating the gravity model based on theoretical advances (Head and Mayer 2014;Yotov, Piermartini, and Larch 2016;Egger, Larch and Yotov 2022), we also estimate the model in its multiplicative form using a Poisson pseudo maximum likelihood (PPML) estimator: where the dyadic fixed effects associated with trade,   , represent the time-invariant characteristics of the trade relationship between i and j, as above.TPijt represents time-variable bilateral factors, such as being a member of the WTO or of an RTA.Finally, εijt is the error term and is assumed to be identically or independently distributed.As before, we extract the country-time fixed effects from Model (10) and use them in a second step, similarly to Equation (9).

Effect of Trade Facilitation on Exports and GVCs
In this section, we estimate a panel data model using country-level variables related to trade and GVCs and link them with TF factors.Estimating the models in first differences will also allow us to infer the effect of the pandemic on fostering the use of electronic documents and procedures to decrease trade costs.This, however, will only be possible with trade variables, given that trade-in value added and the related variables are only available until 2018.
As dependent variables, we alternatively consider total trade, trade in goods, and trade in services.Moreover, we also use the participation of countries in GVCs, using the two proxies described in the data and variables section, that is, the GVC participation index and the GVC position of a country.The empirical strategy to infer the effect of trade facilitation improvements on trade and GVC variables consists in estimating a panel data model that controls for country and time unobserved heterogeneity and estimated with the variables in first differences.This method is also known as a "random trend model" (Wooldridge 2010), which is an extension of the standard unobserved effects models for panel data.The model allows each country to have its own time trend.The country-specific trend is an additional source of heterogeneity.The estimated model is given by: where yit is the natural log of the trade and GVC variables and gi is (roughly) the average growth rate over a period (holding the explanatory variables fixed).Since we would like to allow θi and gi to be arbitrarily correlated with the other explanatory variables, our analysis is within a fixed-effects framework.Our approach to estimating Model ( 11) is to difference away θi and estimate the model given by: ∆ln   =   +  1 ∆  +  3 ∆  + ∆  (12) where we have used the fact that git-gi(t-1) = gi and Equation ( 12) becomes the standard fixed-effects model.In differencing the equation to eliminate θi we lose one time period, so that Equation ( 12) applies to T-1 time periods.We are able to apply fixed-effects methods to Equation ( 12) since our trade facilitation indicators have at least four distinct waves, and the minimum requirement to estimate this model is T = 3.According to Wooldridge (2010), it is reasonable to assume that the first difference of the residuals is serially uncorrelated, in which case the FE method applied to Equation ( 12) is attractive.

Cost of Trading Across Countries
The third set of estimations takes the variables from the World Bank Doing Business data set as dependent variables, one by one.These will serve us to answer the question of whether the implementation of the trade facilitation agreement has indeed reduced the cost to export, cost to import, and the number of documents used for export and import, respectively.A similar model to that in the previous subsection will be used, but without exploiting the panel dimension of the data, given that the time variation is almost not there.Another limitation is that these variables are only available until 2019.Therefore, we will be able to show correlations but not causality in this case.
A similar exercise is done using the different components of the Logistics Performance Index (LPI), which is also available from the World Bank, but only for the years 2016 and 2018.

MAIN RESULTS
Table 2 shows the main results from estimating the gravity model of trade given by Equation ( 7) for trade in manufactured goods. 8The results, including gravity variables instead of dyadic fixed effects, as in Equation ( 6), are shown in the Appendix (Tables A.3 and A.4). Model (7) includes bilateral and time fixed effects, and hence it controls for bilateral unobserved heterogeneity and common time effects.Column (1) reports the result for the trade facilitation index (TFI).The reported coefficient indicates that a one percentage point increase in the index for exporters increases trade by around 1.34%.The overall TF index includes the above-mentioned components (transparency, formalities, paperless, institutions, and cross-border).First, Columns (2) to (6) present the estimates for each component of the specific TF scores.The highest effect in magnitude is obtained for the transparency rating score (1.97), followed by formalities and institutions, whereas paperless TF and transit TF measures are not statistically significant.Second, Columns (7) to (9) present the coefficients for each component of the sustainable TF scores.Only the agricultural component is statistically significant, whereas the TF for SMEs and for women is not.Interestingly, only the TF indicators for the exporter are statistically significant, whereas those for the importer are not.Similarly, only the GDP of the exporter is statistically significant and shows a more than proportional effect on exports, whereas the GDP for the importer is not.With regard to the RTA dummy variable, it shows a weakly significant effect, indicating that exports are around 44% higher when countries belong to the same RTA according to Column (1): [exp(0.368)-1]*100)].Note: Robust standard errors in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1.All columns include country-pair and time fixed effects.Trade facilitation variables are described in Appendix A.1.TFI denotes the trade facilitation index of the exporter/importer country.Id denotes the number of bilateral trade relations.GDP denotes the gross domestic product per capita of the exporter/importer in constant USD of 2017.RTA denotes regional trade agreements and takes the value of one when the trading countries belong to the same regional trade agreement and zero otherwise.
Columns (1) and (2) report the results for the specification with exports in natural logarithms and Columns (3) and ( 4) in levels for exports of manufactured goods and primary products, respectively.The results for the RTA variable for manufactured goods are consistent with the analysis in Table 2 (compare 0.368 with 0.231 in the first columns of Tables 2 and 3, respectively).From these models, the country-time fixed effects (CTFEs) are extracted and used in a second-step estimation, the results of which are shown in Table 4, which presents the outcomes obtained using Column (1) in Table 3 as a first step.The correlation between these CTFEs and the TF index (TFI) is shown in Figure 6, which shows a clear positive correlation.Note: Robust standard errors in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1.Exp_ma and Exp_pri denote total exports in manufactured and primary products, respectively.RTA denotes regional trade agreements and takes the value of 1 when the trading countries belong to the same regional trade agreement and 0 otherwise.The results in Columns ( 1) and ( 2) are obtained with the Stata command reghdfe and in ( 3) and ( 4) with ppml_panel_sg.
Source: Author's elaboration.It, jt, and ij denote exporter-time, importer-time, and bilateral fixed effects.

Figure 6: Scatterplot of Country-Time FE from the Gravity Model and Trade Facilitation Scores
Source: Author's elaboration.Using the ln of the exporter-time FE (lnexpFE on y-axes) from Model (3) in Table 3; tfi_exp denotes the trade facilitation index of the exporter countries.
It is worth noting that the estimates obtained in Table 4 for the TF indicators are similar in general to those in Table 2, but slightly smaller in magnitude.In particular, it is confirmed that transparency measures, formalities, and institutions supporting TF are the most effective, whereas paperless TF is only weakly significant, and transit TF does not show a clear effect.In this case, agriculture-related TF shows a significant positive effect concerning sustainable TF indicators.However, the effect of TF measures related to SMEs, in this case, is negative and significant, perhaps reflecting the small cost firms have to incur to adopt the measures in the short term.Likewise, TF for women does not show a significant effect on trade.Note: Robust standard errors in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1.TFI denotes the overall trade facilitation index (TFI) as defined in Table A1.GDP exporter denotes the gross domestic product per capita of the exporter in constant USD of 2017.All columns include country and time fixed effects.The number of id refers to the pair of countries included in the sample.
Since the effects could differ by income level, Table 5 presents estimates for highincome countries (HICs), low-income countries (LICs), upper-middle-income countries (UMCs), and low-middle-income countries (LMCs).It seems that all income groups benefit from the implementation of transparency TF, formalities TF, and institution TF.
In contrast, paperless TF mainly benefits UMCs and LMCs, and transit TF only LICs.
As regards the sustainable TF measures, agriculture TF is beneficial for all, but not so SMEs and women TF measures.A1 for the definition of the TF variables in the first row.High-income countries (HICs), low-income countries (LICs), upper-middle-income countries (UMCs), and low-middle-income countries (LMCs).

Results from Country-Level Regressions: Trade, GVC, Cost, and Time to Trade
This section presents the results from estimating Equation ( 12) for trade and GVC variables.The estimated coefficients and the corresponding confidence bands at the 5% significance level are shown in Figures 7-11 for total trade, trade in goods, trade in services, GVC participation, and the GVC position index.The coefficients shown in Figure 7 indicate that TF related to transparency, formalities, institutions, and paperless trade explains total exports of goods and services at the aggregated level.However, cross-border procedures and transit at the border TF do not show any clear effect, as is also the case for sustainable TF indicators (SMEs, agriculture, women).When distinguishing between trade in goods (Figure 8) and trade in services (Figure 9), we see that most of the effects are due to trade in goods, whereas only the institutions component is weakly significant for trade in services.Next, Figures ( 10) and ( 11) show the results for the GVC participation and the GVC position index described in the data section.According to the results shown in Figure 10, only the subcomponent referring to institutions for TF seems to have a clear effect on GVC participation, whereas the component cross-border TF is significant at the 10% level.When the GVC position is examined as shown in Figure 11, however, almost none of the TF components show a significant effect at conventional levels, with only institutions for TF showing a 10% significance.Finally, in order to see whether the implementation of TF measures is helping to reduce the time and cost to trade across borders, we show a graphical representation of the results obtained from simple regressions that do not exploit the panel data structure, and hence show correlations rather than causality.Figure 12 shows the results of using three sets of variables for exporters and importers related to trading across borders, namely the cost and time to export and import, and the number of hours needed to fill the documents required to export and import.All these variables are negatively correlated with the TF index (tfi), as indicated in Figure 12.For instance, an increase in the tfi of one percentage point is related to a decrease in the cost involved in border compliance (costbcx) of $323, which doubles the mean value in the sample.Similarly, the same increase in tfi is related to a reduction in the hours needed for documentary compliance for exporting 139 (timedocx), with a sample average of 51.
We also consider, as shown in Figure 13, whether the TF index is correlated with the LPI and its components, that is: quality of trade and transport-related infrastructure; competence and quality of logistics services (lpilogs); ability to track and trace consignments(lpitrak); efficiency of the customs clearance process (lpicus); ease of arranging competitively priced shipments (lpicomp); and frequency with which shipments reach consignees within the time scheduled (lpitime).It can be seen that all the LPI components are positively correlated with the TF index, and these correlations are all statistically significant.This result indicates that improvements in logistics performance are clearly correlated with better trade facilitation performance and will surely improve trade across borders and reduce trade costs.

Figure 13: Dependent Variables: Logistic Performance Index and its Components -Global Sample
Note: lpiall = logistics performance index(lpi): Overall (1 = low to 5 = high); lpiinf = quality of trade and transport-related infrastructure; lpilogs = competence and quality of logistic services; lpitrak = ability to track and trace consignments; lpicus = efficiency of the customs clearance process; lpicomp = ease of arranging competitively priced shipments; lpitime = frequency with which shipments reach consignee within time scheduled.
Source: Author's elaboration.The lines indicate confidence bands at the 5% significance level and the dots indicate estimated coefficients for each of the components of the trade facilitation indicator.

CONCLUSIONS
The combined interlink between economies and the undisputed importance of trading across countries makes the issue of trade facilitation very relevant in the 21st century more so in a world subject to increasing risks and uncertainties related to pandemics and climatic disasters.This paper evaluates the degree of implementation of trade facilitation measures in the world economy and its correlation with some measures of globalization, namely exports of goods and services and participation in GVCs.The main methodology relies on the gravity equation of trade and econometric techniques for panel data sets.With the help of newly collected data covering the pre-and post-pandemic periods, this paper evaluates the relative importance of several developments all directed towards implementing a number of trade facilitation actions that are expected to reduce the cost of trading across borders and to increase trade in goods and services.
The main results from the gravity model indicate that TF measures related to transparency, institutions, and formalities are of utmost importance, whereas sustainable TF actions are still in their infancy, and more data are needed for a proper evaluation of their effectiveness.The TF implementation is related to a reduction in the time and cost to export and import as well as to improvements in logistics performance, which indicates the importance of reducing the time needed to trade, that is, to export and import, with the implementation of targeted policy measures directed towards reducing these times.Moreover, the implementation of TF measures has a significant effect on exports of manufactured goods overall, which is visible for all TFI sub-components.
The implications of the results for developing countries located in the Asia and the Pacific region are manyfold.For instance, the region comprises 12 least-developed countries, of which five are small islands and four are landlocked.For all of them, policies that invest in trade facilitation will contribute more than proportionally to increasing exports of manufactured goods.In particular, measures directed at improving institutional arrangements for border agencies, creating authorized operators, and publishing average release times (transparency) have a more-thanproportional effect on exports.The results from our models indicate that increasing transparency measures will foster exports of manufactured goods more than proportionally for low-income countries.Those include, for example, advanced publication of new trade regulations on the Internet and stakeholders' consultation of new draft regulations.Moreover, policies directed towards improving the quality of institutions will favor the development and deepening of global value chains.Some examples of actions could be the creation of a National Trade Facilitation Committee or similar body or the existence of a clear national institutional arrangement for border agency cooperation.
The main limitation is related to the lack of data in the last two waves of the survey for sustainable trade facilitation measures, which does not allow us to draw conclusions in this respect.Further work should focus on country-specific analysis in the Asia and the Pacific region using firm-level data.This will enable us to disentangle whether the effects of the trade facilitation measures affect differently large and small firms in the region.Note: Robust standard errors in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1.tfi_exp (tfi_imp) denotes trade facilitation measures of the exporter (importer) country; lgdp_exp (lgdp_imp) denotes the natural log of the gross domestic product of the exporter (importer) country; ld is the natural log of the distance between countries; Contig takes the value of 1 for a pair of countries sharing a border, and 0 otherwise.Comlang_off and Comcol take the value of 1 when a pair of countries share an official language or have ever had a colonial relationship, respectively, and 0 otherwise; rta takes the value of 1 when the trading countries are members of a regional trade agreement, and 0 otherwise; wto_o (_d) takes the value of 1 if country i (country j) is a WTO member.

APPENDIX
Source: Author's elaboration.Note: Robust standard errors in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1.tfi_exp (tfi_imp) denotes trade facilitation measures of the exporter (importer) country; lgdp_exp (lgdp_imp) denotes the natural log of the gross domestic product of the exporter (importer) country; ld is the natural log of the distance between countries; Contig takes the value of 1 for a pair of countries sharing a border, and 0 otherwise.Comlang_off and Comcol take the value of 1 when a pair of countries share an official language or have ever had a colonial relationship, respectively, and 0 otherwise; rta takes the value of 1 when the trading countries are members of a regional trade agreement, and 0 otherwise; wto_o (_d) takes the value of 1 if country i (country j) is a WTO member.

Figure 1 :
Figure 1: Trade Facilitation in Asia and the Pacific in 2021

Figure 2
Figure2indicates that countries in Asia and the Pacific generally perform around the global average in some categories, such as transit and transparency, and slightly below concerning border formalities.Otherwise, Asia and the Pacific have a worse performance than the world average in the case of paperless trade and institutions, meaning there is room for improvement.

Figure 2 :
Figure 2: Trade Facilitation in the World in 2021

Figure 3 :
Figure 3: Degree of Implementation of Transparency TF Measures in Asia and the Pacific

Figures 3 -
Figures 3-5 show specific TF scores (notice that the variations go from 0 to 3) for different subregions in Asia and the Pacific and specific TF indicators.Those are related to transparency (Figure3), paperless trade (Figure4), and formalities (Figure5).Figure3indicates that the best score in transparency items is for countries in East Asia, whereas the worst refers to countries in the Pacific subregion.This means that while the transparency TF measures have been fully implemented in the former, they are still in the planning stage or have been partially implemented in the latter region.In regard to paperless TF measures, Figure4shows that there is room for improvement in most subregions concerning the implementation of electronic

Figure 4 :
Figure 4: Degree of Implementation of Paperless TF in Asia and the Pacific

Figure 5 :
Figure 5: Degree of Implementation of Formalities TF in Asia and the Pacific

Figure 7 :
Figure 7: Dependent Variable Trade in Goods and Services -Global Sample

Figure 8 :
Figure 8: Dependent Variable Trade in Goods -Global Sample

Figure 9 :
Figure 9: Dependent Variable Trade in Services -Global Sample (Trade in Technology-Intensive Goods)

Figure 12 :
Figure 12: Dependent Variables: Cost and Time to Trade Across Borders -Global Sample

Table 1 : Summary Statistics Variable Obs Mean Std. Dev. Min Max
Note: SeeTable A.1 for a description of the variables.TFI denotes the trade facilitation index.Source: Author's elaboration.WTO denotes World Trade Organization.