Policy Research Working Paper 10271 Trade Policies and Sea and Air freight The Impact of COVID-19 Lockdowns on Imports and Exports Socrates Majune Angella Faith Montfaucon Macroeconomics, Trade and Investment Global Practice January 2023 Policy Research Working Paper 10271 Abstract This study analyzes how Indonesia’s international trade was imports subject to non-tariff measures requiring physical affected by its own lockdown policies (domestic) and those inspection, testing, and approval processes. External lock- of its trading partners (external) in response to COVID-19. downs, which also had a larger impact on exports relative The study differentiates between sea freight and air freight, to domestic policies, affected sea and air exports evenly. as well as products affected by specific non-tariff measures. Demand factors (specifically, workplace closures and stay- Event-study results show that the decline in imports (which at-home orders) in the partner countries were the drivers were more negatively affected than exports) was mainly of the decline in exports. Enhancing trade facilitation to attributed to external lockdowns, the impacts of which keep goods moving as smoothly as possible, reforming spe- were more pronounced and persistent for imports enter- cific non-tariff measures, and improving customs and other ing Indonesia by air (due to restrictions to international procedures would ensure fewer disruptions from shocks in travel) and imports subject to port-related non-tariff mea- a globally integrated world. sures. Domestic lockdowns adversely affected intermediate This paper is a product of the Macroeconomics, Trade and Investment Global Practice. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at amontfaucon@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 Trade Policies and Sea and Air freight: The Impact of COVID-19 Lockdowns on Imports and Exports∗ Socrates Majune†and Angella Faith Montfaucon‡ JEL Classification : F10, F13, F14, O53 Keywords : Lockdown policies, Indonesia, COVID-19, Sea cargo, Air cargo, Non-tariff Mea- sures ∗ The authors would like to thank Devaki Ghose and Csilla Lakatos for helpful comments, Bayu Agni- maruto for updating the Indonesia non-tariff measures data and Bradley Robert Larson for the COVID-19 data update. This paper has partially benefited from financial support from the Umbrella Facility for Trade trust fund, financed by the governments of the Netherlands, Norway, Sweden, Switzerland and the United Kingdom. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors and do not necessarily represent the views of the World Bank and its affiliated organizations, or those of the Executive Directors or the countries they represent. All errors are our responsibility. † World Trade Organization and University of Nairobi skmajune@uonbi.ac.ke ‡ World Bank amontfaucon@worldbank.org, corresponding author 1 Introduction While coronavirus (COVID-19) is primarily a health crisis, its ramifications extend to social and economic aspects of countries. Trade collapsed and demand and supply shocks became widespread after pandemic-induced measures were imposed globally to contain trans- missions. Supply-side shocks on exports arose from COVID-19 related deaths and infections, and lockdown policies such as border closures, slow clearance of goods and workplace closures that reduced labor participation resulted in a slump in overall firm and country production capacity. Demand shocks then compounded the supply shocks from declines in aggregate demand due to job redundancies, business closures, cautionary spending and investments. These shocks were in turn transmitted and augmented through disruptions in the global supply chains (Hayakawa and Mukunoki, 2021a; Hayakawa and Mukunoki, 2021b). In this context, this paper assesses the trade effect of the pandemic on the largest economy in Southeast Asia, Indonesia. The study investigates the broad effects of lockdown measures (domestic, i.e., imposed by Indonesia, and external, i.e., imposed by Indonesia’s trading partners) on Indonesia’s imports and exports, including the survival of products in import and export markets. Then, we differentiate between sea and air trade as well as assess the role of specific non-tariff measures (NTMs) on the effect of lockdown measures on trade performance in Indonesia. Additional analyses are also conducted to establish the sensitivity of our results to the type of lockdown policy and the effect of lockdowns on export and import prices. Lockdown is a binary composite indicator variable where one indicates the month any of the following measures were imposed by Indonesia and its trading partners: workplace closure, the closure of public transport, stay-at-home requirements, restrictions on internal movement, and international travel controls. This is done using a monthly series of product- by-country data for the period 20 months before and 20 months after January 2020, i.e., May 2018 to August 2021, in survival analysis and event-study designs. For imports, domestic lockdowns represent demand factors and external lockdowns represent supply factors. For exports, the inverse applies. 1 Results from the product-country relationship survival analysis show that the failure rate of exports and imports increased during both domestic and external lockdowns, but dif- ferences between sea and air freight trade were minimal. Notably, imports subject to NTMs had lower survival rates during both domestic and external lockdowns, especially those sub- ject to port of entry restrictions and import approvals. Event-study results show that overall, imports were more negatively affected by lockdown policies than exports, which recovered during the sample period, unlike imports. The decline in imports was mainly attributed to lockdown policies by Indonesia’s trade partners. These affected imports entering Indonesia by air more than those entering by sea, mainly attributed to restrictions on international travel which limited the availability of air cargo during the pandemic. Imports subject to port of entry restrictions and pre-shipment inspections were also more negatively affected, which seem to have worsened the supply side factors for imports. On the other hand, Indonesia’s lockdown policies on imports mainly negatively affected intermediate imports subject to NTMs, especially measures requiring physical inspection and testing (such as certification with Indonesian product standards or Standar Nasional Indonesia (SNI)) and import approval processes. As for exports, domestic lockdowns had a mixed and large impact on air exports, but not sea exports, while external lockdowns affected sea and air exports relatively evenly. These results suggest that supply factors had a larger role in the impact of external lockdowns on imports than demand factors related to the COVID-19 shock. However, NTMs on Indonesia’s imports exacerbated the impact of demand factors for imports. Meanwhile, demand factors (specifically workplace closures which imply lower or no income, and stay at home orders) in the partner countries were the drivers for the slump in exports. Enhanced trade facilitation to keep goods moving as smoothly as possible – including reforming bur- densome non-tariff measures, improving customs and other procedures - would ensure fewer disruptions from shocks in a globally integrated world. These would also ensure that port inspections and procedures that necessitate port of entry restrictions are conducted more 2 efficiently and avoid backlogs and port delays. In this regard, Indonesia has taken several steps to remove some regulations related to measures such as pre-shipment inspections, and introduce a more integrated risk management system in 2021. The impacts of these remain to be seen. Additionally, procedures that require physical inspections of goods and testing and certification and their processes could be improved. Studies providing evidence of the impact of COVID-19 on international trade have been on the increase. They have so far examined the effect of the virus on global value chains neda-Navarrete (Javorcik, 2020; Vidya and Prabheesh, 2020; Egger and Zhu, 2021; Casta˜ et al., 2020; Espitia et al., 2021; Che et al., 2021; Hayakawa and Mukunoki, 2021a), trade policies (Baldwin and Evenett, 2020; Evenett, 2020; Mendoza, 2021; Brenton et al., 2022), uchel et al., 2020; Zhao et al., 2021; export and import flows (Maliszewska et al., 2020; B¨ Minondo, 2021; Hayakawa and Mukunoki, 2021b; Rose et al., 2021;Fang et al., 2022) and COVID-19 lockdown measures (Majune and Addisu, 2021; Hayakawa and Mukunoki, 2021b; Pei et al., 2021; Arenas et al., 2022). These studies have mainly been cross-country aggregate analyses with few exceptions, employing various approaches in their analysis (descriptive statistics, regression model, gravity model, computable general equilibrium model, difference- in-differences, propensity score matching, logistic regression and event-study methodology). Monthly data is prevalent among papers except for Majune and Addisu (2021) who applied weekly-level data. Majune and Addisu (2021) and Arenas et al. (2022) are closely related to our work because they employed an event-study methodology in their analysis. Majune and Ad- disu (2021) found that COVID-19 had an asymmetric effect on Kenya’s international trade: exports increased by an average of 13% while imports dropped by an average of 23% be- tween July 1, 2019, and June 30, 2020.1 The introduction of lockdown measures by trading partners of the Philippines affected its imports more than exports, leading to 7% and 56% 1 The fall in imports was triggered by the disruption in sea cargo trade with countries that imposed lock- down controls. Furthermore, the authors find that lockdowns imposed by Kenya also boosted the country’s exports and dampened imports. 3 monthly average drops in export and import values, respectively, according to Arenas et al. (2022). We closely follow the two studies with a focus on Indonesia and in turn make three contributions to this literature. Our first contribution is that our paper incorporates non-tariff measures (NTMs) in the analysis, allowing the paper more policy relevance. This is highly relevant for Indonesia, as NTM incidence is among the highest in the region, and their use was prevalent in the ı & Montfaucon, 2021). The novel feature of this contribution is the pre-COVID period (Cal` use of the most updated time-varying data on non-tariff measures, an area in trade policy that has not received much attention in this literature, especially due to the lack of reliable data on NTMs. This allows us to link to the timeline of the pandemic and assess how various lockdown policies differently affected import trade of products subject to specific NTMs. Our second contribution is methodological. We employ an event-study in our study. This approach is preferred over impact-evaluation estimators such as the difference-in-differences (DiD) and Propensity Score Matching (PSM), and a gravity-model. The event-study ap- proach addresses concerns of lockdown anticipation. Since East Asia was the first epicenter of COVID-19, most countries in the region started imposing restrictions earlier than the rest of the world (Arenas et al., 2022), which could have triggered the reaction of exporters and importers in Indonesia in anticipation of the imposition of lockdowns in partner countries. The event study also enables us to observe the reaction of international trade to lockdown policies at high frequency (months), which often changed depending on the spread/wave of the virus in a country. Finally, we add to the literature on COVID-19’s impact on trade in a country case study by using more recent and more detailed trade data in a comprehensive analysis of trade for Indonesia, separating impacts of domestic and external lockdowns on trade. This is unlike other studies on Indonesia such as Eschachasthi (2022), Ing and Vadila (2022), and Olivia et al. (2020). Also existing studies on Indonesia, we assess lockdowns in the context of trade by sea and air. Lockdowns disrupted both maritime shipping, which is the main mode 4 of transportation in international trade, and air freight services. With canceled sailings and flights, port delays, and container shortages, COVID-19 measures may have differences in trade transported by sea or air to an archipelago country like Indonesia (see Majune and Addisu, 2021), with implications for trade facilitation measures. The reminder of this paper is organized as follows. The data is described in section 2. The methodology is presented in section 3 with the results discussed in section 4. Finally, section 5 concludes our study. 2 Data 2.1 Lockdown data Lockdown is a binary variable, a composite indicator where one indicates the month any of the following measures were in place: workplace closure, the closure of public trans- port, stay at home requirements, restrictions on internal movement, and international travel controls. Lockdown is twofold: external and domestic. External lockdown refers to when the indicator refers to lockdown policies by Indonesia’s trading partners and domestic lock- down is a dummy where any of the above-mentioned indicators are applied by Indonesia in a particular month. This definition of lockdowns has been adopted from Majune and Addisu (2021) and differs from country to country for the external lockdown since lockdowns were in place at different times. Data for these indicators is from the Oxford COVID-19 Government Response Tracker (Hale et al., 2021). To explain the pattern of lockdown measures, we display the trend of Indonesia’s lock- down Stringency Index and that of selected countries (Papua New Guinea, India, Thailand, Malaysia, Australia, Singapore, the Philippines and Vietnam) between January 1, 2020, and August 31 2021 in Figure 1. The index is constructed by re-scaling the maximum ordinal values of nine indicators (one health measure and eight closures and containment policies) to range between 0 and 100. The scores are then averaged to get a composite index for each 5 country. A score of 0 indicates no restrictions while 100 signals severe implementation of lockdown policies in a country (Hale et al., 2021). The first lockdown measure in Indonesia was on January 18, 2020, when COVID-19 screening of international travelers was introduced. This was followed by the ban on arrivals from some regions on February 5, 2020. Closure of workplaces, total border closure, closure of public transport, restrictions on internal movement and stay-at-home requirements followed in that order from March 15 and they were all operational by April 1. This is depicted by the consistent rise in the stringency index which remained above 70 until May 20, 2020, when the closure of public transport was temporarily lifted. The index has since fluctuated depending on the severity of the COVID-19 wave. Indonesia had the second highest index among the countries in Figure 1, after Australia, by the end of 2020. All the lockdown measures were in operation in 2021 and the index was relatively stable at around 72. Indonesia had the third highest index, after Vietnam and the Philippines, at the end of the sample period on August 31, 2021. Among the neighboring countries, Singapore was the first to impose lockdown measures (January 2), followed by Vietnam (January 25), India (January 26), Papua New Guinea (January 27), Malaysia (January 30), the Philippines (January 31), Australia (February 1) and Thailand (March 6). All the countries started with international travel controls and progressively introduced domestic measures. Singapore’s stringency index was the lowest of all Indonesia’s neighbors at the end of August 2021 because of the relaxation of the conditions on internal movements and public transport. The Philippines, Vietnam and Australia had the highest stringency indices by the end of August 2021, respectively. 2.2 Trade data Our export and import data is a monthly product-country data series obtained from the Indonesia Statistics Authority (BPS Badan Pusat Statistik ). The data ranges from May 2018 to August 2021. Products are classified at the 8-digit Harmonized System (HS) level. 6 Figure 1: COVID-19 Stringency Index of the Philippines and nearby countries in 2020 Source: Author’s compilation using data from Hale et al. (2021) The data also provides trade disaggregated by the port of entry for imports and port of departure for exports. We are thus able to identify sea and air cargo trade. Air transport transferred approximately 28% and 17% of Indonesia’s imports and exports between May 2018 and August 2021, respectively. Sea imports transported about 72% of Indonesia’s imports and 83% of Indonesia’s exports over the May 2018-August 2021 period.2 The trend of exports and imports in Indonesia between May 2018 and August 2021 is shown in Figure 2. Using May 2018 as the base (=100), we note that the growth rate of exports often exceeded that of imports. We also note that both exports and imports have fluctuated and remained below the base period rate before COVID-19. The start of lockdown measures by Indonesia - mild lockdown - in January 2020 led to a 2.5 percentage point growth in exports and 15.8 percentage point decline in imports in February 2020. This suggests that the reaction of imports to domestic lockdown polices was more sudden and 2 Table A1 in the Appendix displays descriptive statistics for our import and export trade data. 7 severer than exports. Figure 2: Percentage growth rate for Indonesia’s exports and imports (May 2018-August 2021) Note: Mild lockdown indicates the month when few restrictions were in place in Indonesia. Severe lockdown is the month when all measures were imposed. Relaxation is the first month when some lockdown measures were lifted Source: Authors’ compilation using BPS data Exports and imports declined in March 2020 upon the imposition of severe lockdown measures: exports declined by 10.5 percentage points while imports decreased by 22.9 per- centage points between May and April 2020. The response of exports and imports to the relaxation of some lockdown measures in late May was instant as shown by the sudden jump in the two graphs around June 2020 in Figure 2. Though fluctuating, both exports and imports have since grown but imports have remained below the base period while the value of exports has consistently exceeded the base period value from March 2021. This could mean that exports have recovered to the pre-COVID level while imports are yet to attain this level. 8 2.3 NTM data We use an updated version of a novel comprehensive NTM database for Indonesia3 built by the World Bank. The data is hand collected through extensive regulatory checks. It has the most disaggregated NTM classification available, i.e., 3-digit MAST level, and the highest frequency available, monthly series. This classification is the most appropriate for policy analysis as it identifies individual measures introduced or modified by each agency. There are over 600 regulations spanning 13 government ministries and agencies. A total of over 60 3-digit NTMs are available in the data set, each with values coded between 0 and 1 to signify when they were in effect on the corresponding products, that is, the data varies at the month-product pair. The starting point of this database is data collection by the Economic Research Institute for ASEAN and East Asia (ERIA) in collaboration with UNCTAD. Then the regulations are traced forward and backward from 2015 and coded accordingly, using various other sources, primarily through government repositories. This data is the most up-to-date data on NTMs and at a high-level of frequency, both of which make the data most relevant for the aim of this paper. As non-tariff measures usually serve a public policy objective, extensive evaluation needs to be made to identify those that may require reform. We focus on four measures in this analysis that previous research has found to be more restrictive to trade than necessary ı et al. (2022) and Cal` and requiring elimination or reform (see Cal` ı and Montfaucon (2021) based on data up to 2018). These measures are pre-shipment inspections (PSI, MAST code C1), mandatory certification with Indonesian product standards or Standar Nasional Indonesia (SNI, MAST code B7), port of entry restrictions (PoE, MAST code C3), and import approvals (IA, MAST code B14). Additionally, three of these measures have been found to have higher tariff ad valorem equivalents compared to similar measures in other countries in ASEAN (Montfaucon et al., 2022). 3 ı et al. (2022), and Cal` Cali et al. (2021), Cal` ı and Montfaucon (2021) also use an older version of this data set. 9 Among these, import approvals affect the highest number of products, followed by port of entry requirements as shown in Figure 3. Pre-shipment inspections have decreased notably in their incidence in recent years, more so in 2021 following key reforms and products for which SNI is required have also remained relatively low. Figure 3: Share of Import Products affected by Specific non-tariff measures a. Pre-shipment inspections (PSI) b. Port of entry restrictions (PoE) c. Product quality standards (SNI) d. Import approvals (IA) Source: World Bank Indonesia NTM Database. 2.4 Survival analysis Trade survival is the likelihood that a trade relationship stays active for a specific period s and Prusa, 2011). Figure 4 displays Kaplan–Meier (KM) export survival plots for (Besedeˇ export and import trade in Indonesia before and under the domestic and external lockdowns. In both cases, exports had higher survival when there was no lockdown (i.e., before January 2020), be it domestic or external. Panel a shows that 38% of export relationships survived beyond the first month of trading when there are no external lockdowns and 37% when lockdown measures are in place in trading partner countries. More than double of exports survive when there were no external lockdowns compared to when lockdowns were in place by the twelfth and twentieth months. Panel b also shows that during domestic lockdowns, exports had lower survival rates. 10 Figure 4: Indonesia’s export and import survival before and under the lockdown Note: Domestic lockdown indicates the months when Indonesia imposed containment measures (from January 2020). External lockdown is the period partner countries imposed lockdowns (from January 2020). No lockdowns are the months before January 2020 Source: Authors’ compilation using BPS data Figure 4 panel c shows that 40% of import relationships survived beyond the first month of trading when there are no external lockdowns in place, compared to 39% when there are external lockdowns. The survival rate after 12 months for imports was 9% when external lockdowns were not in place and 4% when they were present. The difference in survival rate was 6% by the twentieth month. Panel d also shows that imports survived more when no domestic lockdowns were in place. Figure 5 plots the KM survival graph for air and sea export and import survival during the lockdown. Figure 5 panels a and b indicate that sea exports had a slightly higher survival than air exports within the first eight months of external and domestic lockdowns. The survival rates were relatively even afterwards. As for imports, panels c and d indicate a slightly higher survival rate for air imports compared to sea imports during the external 11 and domestic lockdowns. Figure 5: Indonesia’s Air and Sea export and import survival under the lockdown Note: Domestic lockdown indicates the months when Indonesia imposed containment measures (from January 2020). External lockdown is the period partner countries imposed lockdowns (from January 2020) Source: Authors’ compilation using BPS data Figure 6 plots KM graphs for the import survival of products that were exposed to the four NTMs: pre-shipment inspections, SNI, port of entry restrictions, and import approvals. Products that were subject to these NTMs had lower import survival rates compared to those that were not exposed, during both domestic and external lockdowns. The largest differences in import survival were between products exposed and not exposed to port of entry requirements and import approvals. Notably, these are two of the four measures that have a higher incidence. The result for port of entry restrictions is also consistent with port congestion caused by COVID-19 measures. This will be tested more formally in the next section. 12 Figure 6: Indonesia’s import survival by NTM exposure under domestic and external lockdowns Note: Domestic lockdown indicates the months when Indonesia imposed containment measures (from January 2020). External lockdown is the period partner countries imposed lockdowns (from January 2020) Source: Authors’ compilation using BPS data 3 Methodology An event-study approach is employed to establish the effect of COVID-19 lockdown policies on international trade in Indonesia. This approach is preferred over impact-evaluation estimators such as the difference-in-differences (DiD) and Propensity Score Matching (PSM). DiD and PSM compare the outcomes before and after a treatment has been imposed on a control and treatment group. Given that all countries in our sample imposed lockdowns, we do not have a non-treated control group (Majune and Addisu, 2021). Another advantage 13 of the event-study approach is that it addresses concerns of lockdown anticipation. Since East Asia was the first epicenter of COVID-19, most countries started imposing restrictions earlier than the rest of the world, which could have triggered the reaction of exporters and importers in Indonesia in anticipation of the imposition of lockdowns in partner countries. The event-study also enables us to observe the reaction of international trade to lockdown policies at high frequency (months), which often changed depending on the spread/wave of the virus in a country. The event-study model for the effect of external lockdowns is specified as follows: −2 19 ln(Yipgt ) = αt + αit + αig + j =−24 βj 1{tdif fg = j} + j =0 βj 1{tdif fg = j } + γipt + ϵipgt (1) where the dependent variable,Yipgt , is a natural logarithm of bilateral export and import trade (values/quantities) between Indonesia (p) and its trading partners (g ) across products (i) and time in months (t). Seasonality patterns are controlled for by month-time fixed effects (αt ) and product-month fixed effects (αit ). αt captures any secular variation in exports and imports during a month while αit controls for seasonality and inventory stocking patterns associated with high frequency data such as monthly (Bricongne et al., 2012). αig is the product-partner fixed effects that captures any product and partner characteristics that are correlated with lockdowns and trade flows. In general, the three fixed-effects address the problem of endogeneity, particularly unobserved heterogeneity arising from characteristics of partner countries, product characteristics and seasonality of trade. βj is the coefficient indicating the magnitude of the effect of the lockdown on an export or import trade outcome j months before and after it was implemented. tdif fg is the number of months until and after the lockdown was imposed in a partner country. The period j = −1 is a month prior to the lockdown. It is used as the reference category when interpreting coefficients for other periods. 14 All periods j < −1 are utilized as leads to show the anticipatory effect of the lockdown after the start of the pandemic. Countries imposed lockdowns at different times in our sample following the first case of the virus in China. Hence, there is a possibility of anticipatory effects and failure to account for these effects can understate the effect of the lockdown policies (Majune & Addisu, 2021). All periods j > −1 are considered as lags. Although the maximum period of our data is 40 months, we allow our leads to run up to 24 months since some countries delayed imposing lockdown measures (not all countries imposed lockdowns in January 2020, which could have made leads to be 20 months). On the other hand, the maximum lag period is 19 months since the earliest lockdown event for trading partners of Indonesia was in January 2020. γipt is included to control for the effect of domestic lockdowns on the performance of trade. It contains dummies for the months the lockdown was active in Indonesia and zero for the months the lockdown was absent. ϵipgt is the error term. The event-study model for the domestic lockdown is as follows: −2 19 ln(Yipgt ) = αt + αit + αig + j =−20 βj 1{tdif fp = j} + j =0 βj 1{tdif fp = j } + γigt + ϵipgt (2) Definitions of most terms in Equation 2 resemble those of Equation 1 except for tdif fp which is a number of months until and after the lockdown was imposed in Indonesia. It is a binary variable with one indicating months when Indonesia strictly imposed restrictions (from January 2020) and zero otherwise. Therefore, the number of months until and after the lockdown was imposed is uniform across observations. The maximum period of lags, in this case, is 19 months (February 2020-August 2021) and 20 months for the lead period (May 2018-December 2019). γigt is included in Equation 2 to control for the effect of external lockdowns on export and import flows and is coded as 1 for the months when lockdowns were active in Indonesia’s partners and zero for the months when lockdowns were absent. Event-study coefficients of monthly dummies before and after lockdown policies were either imposed by Indonesia or its trading partners from Equation 2 and Equation 1, re- 15 spectively, are plotted in the results section. The coefficients are interpreted as percentage changes having been transformed into elasticities. Coefficients βj were transformed into elas- ticities as follows βj =(exp βj -1)*100 (see Ullah et al. (2021) and Clarke and Tapia-Schythe (2021) for a guide on conducting event-study analysis in Stata). All regressions are done in sub-samples of imports and exports for: overall trade, type of product, type of transportation and products affected by specific non-tariff measures. We later conduct additional analysis to establish the sensitivity of our results to the type of lockdown policy and the effect of lockdowns on export and import prices. 4 Results 4.1 Imports Overall imports We do not find that domestic lockdown policies affected import values and quantities (Figure 7 a) except for the second month following their implementation. Values declined by 171% while quantities slumped by 102% two months after Indonesia put in place lockdown measures. All other changes in import values and quantities cannot be attributed to domestic lockdowns as the coefficients are not statistically significant. Part b of Figure 7 shows that due to external lockdowns, imports dropped persistently, by 11% a month into the lockdown, and continued to deteriorate, reaching a drop of 30% in the twelfth month and 37% by the nineteenth month. Similarly, the quantity of imports fell by 11% a month into external lockdowns, followed by a 36% drop in the twelfth month and 43% decline by the nineteenth month. The average monthly drops in import quantities and values upon imposition of external lockdowns were 32% and 26%, respectively. The results suggest that containment measures by Indonesia’s trading partners had a larger effect on the country’s imports than its own COVID-19 restrictions (given that all the 16 coefficients for the lockdown period, for both import values and quantities, are statistically significant). This implies that supply factors had a larger role in the impact on imports than demand factors related to the COVID-19 shock. Figure 7: Effect of Domestic and External Lockdowns on Imports in Indonesia a. Domestic Lockdown b. External Lockdowns Note: Blue spike means a coefficient is significant at 1%, 5% or 10%. Source: Authors’ compilation using BPS data Effect of Lockdowns on Air and Sea Imports Maritime shipping is central for goods trade both globally and in Indonesia. Lockdowns impacted the availability of labor to unload ships at ports, especially in countries where this is less automated, or raised costs due to increased protective measures for workers. In 2020, Indonesia had the fourth longest average time in port (0.99 days), after the Russian Federation at 1.31 days, Belgium at 1.04 days, and the United States at 1.03 days (UNCTAD, 2021). We assess whether sea and air trade were affected differently by domestic and external lockdowns. Figure 8 a indicates that the domestic lockdown had a trivial effect on air imports as the coefficients are not statistically significant for both value and quantity. The effect of external lockdowns was instant and persistent on air imports, as shown in part b. The value and quantity of air imports shrunk by 11% and 10%, respectively, a month into the lockdown. The value of air imports later declined consistently between the fourth and sixteenth months where the coefficients are significant with a monthly average of 29%. The quantity of air 17 imports also contracted from the fourth month, and lasted up to the tenth month, with a monthly average of 23%. One of the underlying factors may have been that the cancellation of passenger flights due to travel bans limited the availability of air cargo, resulting in increases in the price of air cargo (OECD, 2020). In subsection 4.3, we formally test which specific lockdown measures were the driver in the effects of imports and exports. We find that Indonesia’s lockdown policies did not significantly affect values and quan- tities of imports by sea cargo (Figure 9 part a). A month into the external lockdowns, however, import value declined by 6% (part b). This was followed by a 12% and 16% decline in the fourth and sixth months, respectively. The value of sea imports seem to have since recovered but these changes are not attributable to the external lockdowns. The lower panel shows that import quantities significantly declined in the first 11 months of the external lockdown by a monthly average of 22%. The coefficients are not statistically significant for the period that follows. In general, we do not find evidence that domestic lockdown policies by Indonesia af- fected either the country’s sea or air imports, but external lockdowns affected air imports more negatively than sea imports. This may be due to restrictions on international travel, which mainly take place by air, and were prevalent. Overall, domestic lockdowns mainly affected Indonesia’s intermediate goods affected by NTMs. External lockdowns mainly led to the decline of imports. Intermediate imports exposed to NTMs We focus this section on intermediate goods imports, as these play a critical role in the access of input varieties. Imported inputs raise productivity, through greater comple- mentarity of inputs and technology/quality transfer (Bas & Strauss-Kahn, 2014). While lockdown policies may have limited imports generally, non-tariff measures on intermediate products may have further worsened the impact of the COVID-19 shock on goods subject to such measures. For instance, limits on mobility of people may have affected processes 18 Figure 8: Effect of Domestic and External Lockdowns on Air Imports a. Domestic Lockdown b. External Lockdowns Note: Blue spike means a coefficient is significant at 1%, 5% or 10%. Source: Authors’ compilation using BPS data Figure 9: Effect of Domestic and External Lockdowns on Sea Imports a. Domestic Lockdown b. External Lockdowns Note: Blue spike means a coefficient is significant at 1%, 5% or 10%. Source: Authors’ compilation using BPS data such as physical inspections of goods and testing and certification. We assess how lockdown policies may have affected intermediate goods imports differently based on whether or not the products are subject to any non-tariff measure and to specific measures. Figure 10 plots the coefficients of the effect of domestic lockdowns on intermediate imports that were exposed to NTMs. We find that domestic lockdowns significantly reduced imports of products that were exposed to NTMs: the average monthly fall was 1,512% for values and 1,058% for quantity between the first and seventh months. Part b shows that the effect of external lockdowns was also negative but sporadic. Import values significantly declined by 11%, a month into the lockdown, then 20% and 22% in the sixth and seventh 19 months, respectively, and 29% in the ninth month. Import quantities declined by 9% after the first month and later by averages of 20% and 28% in the fourth-seventh month and ninth-tenth month, respectively. Figure 11 plots coefficients of the effect of domestic and external lockdowns on inter- mediate imports that were not exposed to any NTM. In sharp contrast to products exposed to NTMs, domestic lockdowns had a delayed and even positive effect on the value of imports that were not affected by NTMs, which increased by an average of 1,655% between the twelfth and eighteenth months of the domestic lockdown. Quantities were not affected. External lockdowns had a negative effect in the initial 11 months on both import values and quantities of products not exposed to any NTM (Figure 11b), rather symmetrically. The upper panel shows that import values significantly declined by 12% in the first month of the external lockdown. They later contracted significantly by a monthly average of 27% between the third and eleventh months. Part b also indicates that the quantity of imports that were not exposed to NTMs declined by 11% in the first month of external lockdowns. They later de- clined (significantly) by a monthly average of 32% between the third and eleventh months. Overall, while NTMs exacerbated the impact of domestic lockdowns, external lockdowns negatively affected imports of products subject and not subject to NTMs indiscriminately. Figure 10: Effect of Domestic and External Lockdowns on Intermediate Imports exposed to any NTM a. Domestic Lockdown b. External Lockdowns Note: Blue spike means a coefficient is significant at 1%, 5% or 10%. This includes all NTMs. Source: Authors’ compilation using BPS data 20 Figure 11: Effect of Domestic and External Lockdowns on Intermediate Imports not ex- posed to any NTM a. Domestic Lockdown b. External Lockdowns Note: Blue spike means a coefficient is significant at 1%, 5% or 10%. This includes all NTMs. Source: Authors’ compilation using BPS data Given that trade is heterogeneously affected by NTMs, in order to separate discussion between more legitimate NTMs and problematic NTMs, we next investigate the role of four specific NTMs in the impact of lockdowns on import values. These are pre-Shipment inspection (PSI), requirement to pass through a specific port (PoE), product quality or performance requirement (SNI), and import approval (IA). Figure 12 shows the effect of external and domestic lockdowns on intermediate goods imports subject to specific NTMs. Domestic lockdowns had a negative and significant effect on intermediate imports subject to import approvals for the entire period of the lockdown. The average monthly decline was 25%. For products required to pass through specific ports of entry, the effect was only negative and significant in the sixth month (-90%), while PSI’s effect was negative and significant in the first, third and ninth months (ranging from -900% to -1,000%). SNI reduced intermediate imports by 21% in the first month and later by an average of 14% between the third and eighth months of the domestic lockdown. External lockdowns affected intermediate imports that were exposed to different NTMs negatively but sporadically. For instance, PSI (code C1) significantly reduced imports in the first two months and in the ninth and tenth months of the external lockdown. The effect of port of entry restrictions (PoE) was negative and significant between the seventh and 21 eleventh months of the lockdown, and we do not find any evidence that SNI and import approvals significantly affected intermediate imports during the external lockdown. This suggests that port-related measures were more relevant in the supply side of imports. Figure 13 shows that products which were not exposed to port of entry restrictions were not significantly affected by the domestic lockdown. There is evidence that absence of PSI and SNI, though sporadic, positively affected intermediate imports during the domestic lockdown. For instance, products which were not exposed to SNI grew by 183% in the first month and later by an average of 319% between the seventh and tenth months of the domestic lockdown. Products which were not exposed to import approval significantly declined during the first three months of the domestic lockdown, a much shorter period than for products subject to import approvals. Consistent with the overall impact of external lockdowns on imports, products which were not exposed to import approvals significantly declined for the entire period of external lockdowns, by a monthly average of 31%. Those that were not exposed to PoE and SNI also declined significantly during the external lockdown except for the second month: the monthly average (third-eleventh month) drop for imports that were not subject to PoE and SNI was 31% and 34%, respectively. Imports which were not exposed to PSI contracted in the first and fourth months by 11% and 23%, respectively. These results suggest that while overall imports were not severely affected by domestic lockdowns by Indonesia, NTMs exacerbated demand-side factors on imports. This was especially the case for imports subject to product certification and approval procedures (i.e., import approval and certification of SNI), which require more physical presence. For instance, compliance with national standards (SNI) is mandatory in Indonesia for thousands of goods, including also domestically produced ones. As certification requires a visit to the factory premises by an Indonesian certifying agency, the cost is considerably higher for imported goods and travel was especially difficult during lockdowns. Only a few agencies, mandated by the Ministry of Industry, can certify a specific product, so producers in need 22 Figure 12: Effects of external and domestic lockdowns on intermediate imports that were exposed to NTMs Note: PSI=Pre-Shipment Inspection; PoE=Requirement to pass through specific port; SNI=Product quality or performance requirement; and IA=import approval; Source: Authors’ compilation using BPS data of certification must often wait a long time to find an agency available for certification. Similarly, obtaining import approval can be cumbersome and its duration often un- predictable, thus raising the costs of importing. The main hurdle in the process is the requirement of recommendation letters from sectoral ministries, who have relatively high discretion in deciding on the recommendation, including whether or not to grant it, how long it takes them to respond and the quantity allowed for the specific approval, which can differ from what the producer requested. Since these processes already come with high uncer- 23 Figure 13: Effects of external and domestic lockdowns on intermediate imports that were not exposed to NTMs Note: PSI=Pre-Shipment Inspection; PoE=Requirement to pass through specific port; SNI=Product quality or performance requirement; and IA=import approval; Source: Authors’ compilation using BPS data tainty, coupled with the uncertainty and challenges of domestic lockdown, the lower demand for intermediate products needed by these products is expected. In late 2021, to ease and simplify the process, the government enacted some policies to remove the recommendation letters for some products. On the other hand, port-related procedures were more relevant in having more adverse impact of external lockdowns (supply side). This may be because these measures more di- rectly impact exporters to Indonesia (for instance, pre-shipment inspection where inspection 24 takes place at the port before products are exported to Indonesia), thus making lockdowns in the exporting countries more relevant. 4.2 Exports Overall exports The effect of domestic and external lockdown policies on Indonesia’s export trade is presented in Figure 14. In general, exports were more affected by external lockdowns than the domestic lockdown policy, but recovered after the eighth month. As shown in the upper panel of part a, domestic lockdowns did not have a significant effect on export values except in the fifteenth month (a 64% growth). The lower panel depicts several instances where the domestic lockdown significantly affected export quantities: they grew by 244% in the third month and by a monthly average of 82% between the twelfth and fifteenth months. The discrepancy between values and quantities signifies the price/exchange rate effect on exports. Part b of Figure 14 shows that export values and quantities significantly declined by 7.5% and 5.7% between the third and eighth months respectively, but recovered from a contraction after the first eight months of the implementation of lockdown policies in destination markets. The initial drop in exports is consistent with Pimenta et al. (2021), who find that lockdown stringency in destination markets reduced exports at the onset of COVID-19 for firms in Portugal. Exports later grew by 9% (for values) and 11% (for quantities) between the tenth and eighteenth months, suggesting some level of adaptation. Effect of Lockdowns on Sea and Air Exports The value of air exports initially improved by a monthly average of 241% in the first five months of the domestic lockdown (Figure 15). This was followed by periods of contraction, particularly after the thirteenth month, when coefficients were statistically significant, with a monthly average of -240%. The effect of the domestic lockdown was not statistically 25 Figure 14: Effect of Domestic and External Lockdowns on Exports in Indonesia a. Domestic Lockdown b. External Lockdowns Note: Blue spike means a coefficient is statistically significant Source: Authors’ compilation using BPS data significant on quantities of air exports in its first 12 months, but significantly reduced them by 129% and 65% in the thirteenth and fifteenth months, respectively, and later improved them by 185% in the sixteenth month. This may signal large variation in prices as the effects are initially seen in values but not quantities. We test this more formally in the next section. External lockdowns reduced air export values immediately they were imposed at a monthly average of 9% in the first three months. The value of air exports later improved from the ninth month with a monthly average of 31%. Air export quantities declined by 7% and 8% in the first and second months of the external lockdown, respectively. They later grew by a monthly average of 25% from the tenth month. In general, domestic and external lockdowns have had an inverse effect on air exports, with the impact of domestic lockdowns having a much larger magnitude. The domestic lockdown boosted air exports at its initial stage, while restrictions by partner countries dampened it. However, air exports improved at the latter stages of external lockdowns but declined from the domestic containment measures. We do not find evidence that domestic lockdowns affected exports by sea cargo, which transported approximately 83% of Indonesia’s exports in the May 2018-August 2021 period (Figure 16a). Figure 16b shows that external lockdowns had a brief disruption on sea export values. They shrunk by a monthly average of 4.6% between the fourth and sixth months 26 but later grew by a monthly average of 12% between the ninth and nineteenth months. The quantity of sea exports grew in the second month, then grew by a monthly average of 16% between the ninth and nineteenth months. Figure 15: Effect of Domestic and External Lockdowns on Air Exports a. Domestic Lockdown b. External Lockdowns Note: Blue spike means a coefficient is statistically significant Source: Authors’ compilation using BPS data Figure 16: Effect of Domestic and External Lockdowns on Sea Exports a. Domestic Lockdown b. External Lockdowns Note: Blue spike means a coefficient is statistically significant Source: Authors’ compilation using BPS data 4.3 Mechanism Type of Lockdown Policy Lockdown measures constituted various types of policies. Since the analysis shows that external lockdowns affected Indonesia’s international trade more than domestic measures, we 27 test the sensitivity of our results to the different types of measures put in place by Indonesia’s trade partners. We assume that different measures adequately capture the severity of the impact of the lockdown restrictions on trade flows as opposed to a compound indicator. To do this, we test the effect of workplace closure, stay-at-home requirements, restrictions on internal movement, and international travel controls on Indonesia’s bilateral export and import values. Each indicator is a dummy variable for the months that it was active in a partner country. We assess the average effect of lockdown indicators for all the months that they were active (overall) and at intervals of 1-4 months, 5-8 months, 9-12 months, 13-16 months and 17-19 months. This gives a glimpse of their effects over time. The top panel of Table 1 indicates that the closure of workplaces had the most severe effect on exports: the overall effect was -3.6%, which is the average over the study period. Since this is the closure of workplaces in Indonesia’s trading partners, it suggests that a slump in demand in partner countries - potentially from unemployment - was the cause for low demand of products from Indonesia. The negative effect was acute in the first four months and from the 17-19th months. Stay-at-home requirements by Indonesia’s partners also reduced Indonesia’s exports: the average monthly drop was 1.6%, mainly driven by the first 12 months when the measures were in place. The overall effect of restrictions on internal movements and international travel on exports is positive although they were negative in the first eight months of trading. The lower panel shows that the overall effect of all lockdown measures on imports was negative, signaling supply-side disruptions. Restrictions on international travel by trading partners by far had the largest negative impact on Indonesia’s imports, with an overall effect of -23%, and continued to worsen over time. Cancellation of passenger flights linked to travel restriction limited the availability of air cargo during the pandemic (OECD, 2020), and this result is consistent with imports by air being more affected by external lockdowns compared to sea cargo. The overall effects of workplace closure and stay-at-home requirements are -9.6% and -8.8%, respectively. Furthermore, the coefficients are large for the first 12 months, 28 implying that disruptions in production reduced imports from countries which had these measures early on during the pandemic. Table 1: Average event-study coefficients for Indonesia’s exports and imports of by external lockdown indicators Exports Period Workplace Stay-at-home Internal movement International travel Overall -5.613 -1.872 3.889 2.697 1-4 months -5.688 -5.778 -5.142 -7.227 5-8 months -4.379 -4.872 -6.794 9-12 months -4.180 -4.087 8.793 6.267 13-16 months -5.015 7.250 8.016 10.617 17-19 months -8.805 x x 10.624 Imports Overall -9.974 -8.683 -2.812 -24.005 1-4 months -18.906 -21.535 -16.768 -9.567 5-8 months -22.226 -18.412 -18.033 -20.651 9-12 months -13.876 -13.134 -11.706 -25.121 13-16 months -8.222 -8.688 9.039 -28.598 17-19 months 13.360 18.353 23.409 -36.088 Note: All coefficients are significant at 10% level. x indicates that all coefficients are not significant in a specific period. Source: Authors’ compilation Import and export prices The second mechanism we test entails establishing the effects of domestic and external lockdowns on export and import prices. This will further attempt to show differences in quantities and values that some results may have shown. Prices are calculated as the ratio of value to quantity at the product-partner-month level for both exports and imports. As before, coefficients are transformed into elasticities. Effects of the domestic lockdowns on import prices are shown in the fourth column of Table A2 (only coefficients for lags are presented due to space constraint). Import prices significantly declined from the third month to the eighteenth month by a monthly average of 316% due to Indonesia’s lockdown policies. As we did not find that domestic lockdowns 29 Table 2: Average event-study coefficients for Indonesia’s exports and imports of by domestic lockdown indicators Exports Period Workplace Stay-at-home Internal movement International travel Overall 84.930 143.418 x 11.173 1-4 months x 189.100 x x 5-8 months x 145.008 x x 9-12 months x 96.147 x x 13-16 months 84.930 x x 11.173 17th month x x x x Imports Overall -0.577 -15.350 -0.494 -0.480 1-4 months -21.042 -31.525 -20.965 -20.965 5-8 months -10.889 -18.486 -10.751 -10.750 9-12 months 4.524 -8.269 4.561 4.643 13-16 months 13.407 -3.121 13.501 13.484 17th month 11.112 x 11.184 11.189 Note: All coefficients are significant at 10% level. x indicates that all coefficients are not significant in a specific period. Source: Authors’ compilation significantly affected import trade except for intermediate goods imports subject to NTMs, this may signal that lower demand for imports may have led to lower prices. The external lockdown led to an increase in import prices, specifically between the third and tenth months when prices significantly increased by a monthly average of 6.2%. This suggests that lower supply from the origin countries due to lockdowns may have led to higher prices. We find no evidence of external lockdowns affecting import prices after the tenth month. As displayed in Table A2, the third column shows that export prices contracted imme- diately after Indonesia imposed lockdown measures and this effect consistently lasted for the first 12 months of the lockdown. The effect started with a 392% decline in the first month and consistently remained negative until the twelfth month (the average monthly contraction in export prices between the first and twelfth months was 249%). Export prices also shrunk in the fourteenth month by 61%. The second column shows that export prices significantly declined by an average of 2.3% in the fourth and fifth months following external lockdowns. 30 This was followed by a 2.2% drop in the seventh month and later by a monthly average of -3.8% between the tenth and sixteenth months. In general, export prices decreased more due to domestic than external lockdowns in the first 12 months, going by the magnitude of the coefficient and starting period of the effect. Nonetheless, the effect of external lockdowns on export prices was long-lived as shown by the 4% decline in the nineteenth month. The decrease in export prices due to external lockdowns may also signal that lower demand may have led to lower price setting by Indonesian exporters. 5 Conclusion This study sought to analyze how Indonesia’s international trade was affected by its own lockdown policies (domestic) and those of its trading partners (external) in response to COVID-19, differentiating between sea and airports as well as imports affected by non-tariff measures (NTMs). We employed survival analysis and event-study approaches on monthly- product-country export and import data ranging from May 2018 to August 2021. Results from the survival analysis show that the failure rate of exports and imports increased during both domestic and external lockdowns, but sea exports had higher survival than air exports, while air had higher import survival rates. Imports subject to NTMs also had lower survival rates during both domestic and external lockdowns, especially those subject to port of entry restrictions and import approvals. Event-study results show that overall, imports were more negatively affected by lockdown policies than exports, which recovered during the sample period, unlike imports. The decline in imports was mainly attributed to lockdown policies by Indonesia’s trade partners, which affected all imports. The impact of external lockdowns was more pronounced and persistent for imports entering Indonesia by air, while imports by sea recovered relatively faster. We find no impact of Indonesia’s lockdown policies on imports, except for negatively affecting intermediate imports subject to non-tariff measures. 31 While external lockdowns also had a larger impact on exports than domestic policies, exports recovered past the eighth month of both domestic and external lockdowns. Domestic lockdowns had a mixed and large impact on air exports, but not sea imports, while external lockdowns affected sea and air exports relatively evenly. The result implies that supply factors (specifically restrictions on international travel, workplace closures, stay-at-home orders and internal movement restrictions, in that order) had a larger role on the impact on imports than demand factors related to the COVID-19 shock. However NTMs exacerbated the impact of demand factors for imports. Meanwhile, demand factors (specifically workplace closures and stay-at-home orders) in the partner countries were responsible for the slump in exports. Enhanced trade facilitation to keep goods moving as smoothly as possible – including identifying unnecessary non-tariff measures, customs procedures with limited human inter- vention such as integrated risk management, and improving processes on necessary non-tariff measures of specific goods, would ensure fewer disruptions from shocks in a globally inte- grated world. These would also ensure that port inspections and procedures that necessitate port of entry restrictions are conducted more efficiently and avoid backlogs and port delays. In this regard, Indonesia has removed several regulations on measures such as pre- shipment inspections,4 to only apply to a limited number of products; and introduced reg- ulations to improve risk management between customs and the Ministry of Trade.5 Nev- ertheless, procedures on import approval and port of entry requirements still affect a large share of imports, and certification for national standards (SNI), while affecting a smaller share of goods, remains quite costly and may warrant further reform. These procedures also require physical inspections of goods and testing and certification and their processes could be improved. 4 In 2021, Indonesia revoked over 20 regulations on PSI 5 In 2021, Indonesia enacted a law to improve information sharing, including procedures that would enable traders with priority status with customs to also be recognized by Ministry of Trade. 32 References Arenas, G. C., Majune, S., & Montfaucon, A. F. (2022). The impacts of lockdown poli- cies on international trade in the philippines. https : / / openknowledge . worldbank . org / bitstream / handle / 10986 / 36882 / The - Impacts - of - Lockdown - Policies - on - International-Trade-in-the-Philippines.pdf?sequence=1&isAllowed=y Baldwin, R., & Evenett, S. (2020). Covid-19 and trade policy: Why turning inward won’t work. Centre for Economic Policy Research (CEPR). https://voxeu.org/content/ covid-19-and-trade-policy-why-turning-inward-won-t-work Bas, M., & Strauss-Kahn, V. (2014). Does importing more inputs raise exports? firm-level evi- dence from france. Review of World Economics (Weltwirtschaftliches Archiv), 150 (2), 241–275. https://doi.org/10.1007/s10290-013-0175-0 s, T., & Prusa, T. J. (2011). The role of extensive and intensive margins and export Besedeˇ growth. Journal of development economics, 96 (2), 371–379. https://doi.org/10.1016/ j.jdeveco.2010.08.013 Brenton, P., Ferrantino, M. J., & Maliszewska, M. (2022). Reshaping global value chains in light of covid-19: Implications for trade and poverty reduction in developing countries. World Bank Publications. https://openknowledge.worldbank.org/bitstream/handle/ 10986/37032/9781464818219.pdf?sequence=5 e, L., Gaulier, G., Taglioni, D., & Vicard, V. (2012). Firms and Bricongne, J.-C., Fontagn´ the global crisis: French exports in the turmoil. Journal of international Economics, 87 (1), 134–146. https://doi.org/10.1016/j.jinteco.2011.07.002 uchel, K., Legge, S., Pochon, V., & Wegm¨ B¨ uller, P. (2020). Swiss trade during the covid-19 pandemic: An early appraisal. Swiss journal of economics and statistics, 156 (22), 1– 15. https://doi.org/10.1186/s41937-020-00069-3 Cali, M., Le Moglie, M., & Presidente, G. (2021). Gain without Pain ? Non-Tariff Measures, Plants’ Productivity and Markups (Policy Research Working Paper Series No. 9654). The World Bank. https://ideas.repec.org/p/wbk/wbrwps/9654.html 33 ı, M., Ghose, D., Montfaucon, A. F. L., & Ruta, M. (2022). Trade policy and exporters’ re- Cal` silience: Evidence from indonesia (Policy Research Working Paper Series No. 10068). The World Bank. https://EconPapers.repec.org/RePEc:wbk:wbrwps:10068 ı, M., & Montfaucon, A. F. (2021). Non-Tariff Measures, Import Competition, and Ex- Cal` ports (Policy Research Working Paper Series No. 9801). The World Bank. https : //ideas.repec.org/p/wbk/wbrwps/9801.html neda-Navarrete, J., Hauge, J., & L´ Casta˜ omez, C. (2020). Covid-19’s impacts on global opez-G´ value chains, as seen in the apparel industry. Development Policy Review. https : //doi.org/10.1111/dpr.12539 Che, Y., Liu, W., Zhang, Y., & Zhao, L. (2021). China’s exports during the global covid-19 pandemic. Frontiers of Economics in China, 15 (4), 541–574. https : / / doi . org / 10 . 3868/s060-011-020-0023-7 Clarke, D., & Tapia-Schythe, K. (2021). Implementing the panel event study. The Stata Journal, 21 (4), 853–884. https://doi.org/10.1177/1536867X211063144 Egger, P. H., & Zhu, J. (2021). How covid-19 travels in-and outside of value chains and then affects the stock market: Evidence from china. The World Economy. https : //doi.org/10.1111/twec.13134 Eschachasthi, R. (2022). Exporters in the time of covid-19 pandemic: Evidence from indone- sia. Economics and Finance in Indonesia, 68 (1), 1–16. http://efi.ui.ac.id/index.php/ efi/article/view/945 Espitia, A., Mattoo, A., Rocha, N., Ruta, M., & Winkler, D. (2021). Pandemic trade: Covid- 19, remote work and global value chains. The World Economy. https://doi.org/10. 1111/twec.13117 Evenett, S. J. (2020). Sicken thy neighbour: The initial trade policy response to covid-19. The World Economy, 43 (4), 828–839. https://doi.org/10.1111/twec.12954 Fang, J., Ou, J., & Yao, S. (2022). On covid-19 pandemic and china’s foreign trade. The World Economy. https://doi.org/doi.org/10.1111/twec.13269 34 Hale, T., Angrist, N., Goldszmidt, R., Kira, B., Petherick, A., Phillips, T., Webster, S., Cameron-Blake, E., Hallas, L., Majumdar, S., et al. (2021). A global panel database of pandemic policies (oxford covid-19 government response tracker). Nature human behaviour, 5 (4), 529–538. https://doi.org/10.1038/s41562-021-01079-8 Hayakawa, K., & Mukunoki, H. (2021a). Impacts of covid-19 on global value chains. The Developing Economies. https://doi.org/10.1111/deve.12275 Hayakawa, K., & Mukunoki, H. (2021b). Impacts of lockdown policies on international trade. Asian Economic Papers, 73–91. https://doi.org/10.1162/asep a 00804 Ing, L. Y., & Vadila, Y. (2022). Covid-19: Impacts of indonesia’s trade (discussion paper 415). ERIA. https://www.eria.org/uploads/media/discussion-papers/FY21/Covid- 19-Impacts-on-Indonesias-Trade.pdf Javorcik, B. (2020). Reshaping of global supply chains will take place, but it will not happen fast. Journal of Chinese Economic and Business Studies, 18 (4), 321–325. https:// doi.org/10.1080/14765284.2020.1855051 Majune, S., & Addisu, L. (2021). The effect of lockdown policies on international trade flows from developing countries: Event study evidence from kenya (working paper 148). Brookings Institution. https://www.brookings.edu/wp-content/uploads/2021/03/ The-effects-of-Lockdown-Policies-Kenya.pdf Maliszewska, M., Mattoo, A., & Van Der Mensbrugghe, D. (2020). The potential impact of covid-19 on gdp and trade: A preliminary assessment (policy research working paper 9211). The World Bank Group. https://openknowledge.worldbank.org/bitstream/ handle/10986/33605/The-Potential-Impact-of-COVID-19-on-GDP-and-Trade-A- Preliminary-Assessment.pdf Mendoza, A. (2021). Disruptions in global value chains due to covid-19: Stylized facts and policy lessons. Philippine Review of Economics, 57 (2), 214–240. https://doi.org/10. 37907/9ERP1202JD 35 Minondo, A. (2021). Impact of covid-19 on the trade of goods and services in spain. Applied Economic Analysis, 29 (85), 58–76. https://doi.org/10.1108/AEA-11-2020-0156 Montfaucon, A. F., Khan, S. Y., & Agnimaruto, B. (2022). The relative cost of non-tariff measures in indonesia: A benchmarking exercise with other asean economies [unpub- lished]. OECD. (2020). Oecd policy responses to coronavirus (covid-19) - covid-19 and international trade: Issues and actions. OECD. https : / / www . oecd . org / coronavirus / policy - responses/covid-19-and-international-trade-issues-and-actions-494da2fa Olivia, S., Gibson, J., & Nasrudin, R. a. (2020). Indonesia in the time of covid-19. Bulletin of Indonesian economic studies, 56 (2), 143–174. https://doi.org/10.1080/00074918. 2020.1798581 Pei, J., de Vries, G., & Zhang, M. (2021). International trade and covid-19: City-level evidence from china’s lockdown policy. Journal of Regional Science. https://doi.org/10.1111/ jors.12559 Pimenta, A. C., Gouveia, C. M., & Amador, J. (2021). Covid-19, lockdowns and international trade: Evidence from firm-level data (Working Papers w202114). Banco de Portugal, Economics and Research Department. https://www.bportugal.pt/sites/default/files/ anexos/papers/wp202114.pdf Rose, A., Walmsley, T., & Wei, D. (2021). Spatial transmission of the economic impacts of covid-19 through international trade. Letters in Spatial and Resource Sciences, 1–28. https://doi.org/10.1007/s12076-021-00271-8 Ullah, S., Zaefarian, G., Ahmed, R., & Kimani, D. (2021). How to apply the event study methodology in stata: An overview and a step-by-step guide for authors. https://doi. org/10.1016/j.indmarman.2021.02.004 UNCTAD. (2021). The review of maritime transport 2021. United Nations Conference on Trade; Development (UNCTAD). https://unctad.org/system/files/official-document/ rmt2021 en 0.pdf 36 Vidya, C., & Prabheesh, K. (2020). Implications of covid-19 pandemic on the global trade networks. Emerging Markets Finance and Trade, 56 (10), 2408–2421. https://doi.org/ 10.1080/1540496X.2020.1785426 Zhao, Y., Zhang, H., Ding, Y., & Tang, S. (2021). Implications of covid-19 pandemic on china’s exports. Emerging Markets Finance and Trade, 57 (6), 1716–1726. https:// doi.org/10.1080/1540496X.2021.1877653 37 Appendix A1 Air export and import survival Figure A1: Indonesia’s air export and import survival before and under the lockdown Note: Domestic lockdown indicates the months when Indonesia imposed containment measures (from January 2020). External lockdown is the period partner countries imposed lockdowns (from January 2020). Source: Authors’ compilation using BPS data 38 A2 Sea export and import survival Figure A2: Indonesia’s air export and import survival before and under the lockdown Note: Domestic lockdown indicates the months when Indonesia imposed containment measures (from January 2020). External lockdown is the period partner countries imposed lockdowns (from January 2020). Source: Authors’ compilation using BPS data 39 Table A1: Descriptive Statistics Variable Obs. Mean Std. Dev. Min Max Exports Data Export value (overall) 1,476,300 403,687 5,023,029 0.001 711,000,000 Export value (air) 452,106 225,616 2,662,157 0.001 478,000,000 Export value (sea) 1,023,245 482,433 5,765,952 0.001 711,000,000 Export quantity (overall) 1,476,300 1,398,234 67,700,000 0.001 12,900,000,000 Export quantity (air) 452,106 21,489 387,928 0.001 165,000,000 Export quantity (sea) 1,023,245 2,007,507 81,400,000 0.001 12,900,000,000 Imports Data Import value (overall) 2,346,615 240,524 2,790,213 1 441,000,000 Import value (air) 1,358,330 114,923 1,250,980 1 272,000,000 Import value (sea) 985,151 411,183 4,038,814 1 441,000,000 Import quantity (overall) 2,346,615 233,452 5,756,140 1 1,520,000,000 Import quantity (air) 1,358,330 20,456 848,799 1 490,000,000 Import quantity (sea) 985,151 526,239 8,815,848 1 1,520,000,000 Note: Values for exports and imports by road are not shown due to their negligible shares (less than 1% of total trade) Source: Authors’ compilation using BPS data 40 Table A2: Effects of domestic and external lockdown policies on export and import prices Exports Imports Lockdown event dummies External Domestic External Domestic Lag0 -0.126 -392.293** -0.383 -34.943 (0.734) (164.349) (0.726) (21.381) Lag1 -0.038 -346.369** 0.593 - (0.842) (155.041) (1.048) - Lag2 -0.944 -340.488** 1.550 -50.060 (0.975) (145.363) (1.419) (40.373) Lag3 1.252 -341.004** 4.486** -98.423* (1.105) (140.669) (1.784) (58.602) Lag4 -2.322** -291.985** 7.022*** -137.802** (1.170) (130.409) (2.151) (69.642) Lag5 -2.220* -256.019** 7.176*** -162.927** (1.262) (124.485) (2.498) (81.428) Lag6 -1.173 -273.459** 6.173** -189.916* (1.272) (113.473) (2.849) (97.613) Lag7 -2.196* -267.951** 5.953* -225.947** (1.322) (104.200) (3.199) (113.532) Lag8 -1.128 -216.250** 6.306* -211.697* (1.368) (92.074) (3.548) (123.501) Lag9 -0.758 -176.000** 6.452* -286.061** (1.400) (84.331) (3.902) (138.343) Lag10 -2.853** -183.745** 7.354* -327.974** (1.441) (75.592) (4.256) (153.803) Lag11 -4.277*** -154.806** 4.827 -348.036** (1.473) (65.360) (4.612) (167.183) Lag12 -4.516*** -129.789** 5.778 -381.294** (1.500) (52.542) (4.971) (181.223) Lag13 -3.615** -48.062 5.369 -409.336** (1.514) (36.802) (5.333) (193.804) Lag14 -3.447** -60.840** 7.691 -394.609* (1.548) (30.092) (5.701) (205.017) Lag15 -4.047** -19.521 5.645 -449.162** (1.585) (23.305) (6.067) (227.685) Lag16 -3.331** 0.000 5.907 -453.426** (1.646) (0.000) (6.444) (230.057) Lag17 -0.492 4.818 4.860 -473.137** (1.751) (26.875) (6.821) (239.356) Lag18 -2.726 62.052 4.949 -525.217* (1.871) (64.290) (7.213) (279.400) Lag19 -3.979* 106.477 5.262 - (2.108) (67.731) (7.608) - Constant 176.526*** 515.447*** 281.618*** 446.139*** (2.736) (189.011) (4.859) (63.718) Observations 1,467,415 1,467,415 1,879,798 1,879,798 Number of pair id 150,489 150,489 184,744 184,744 R-squared 0.002 0.002 0.001 0.001 Note: Estimation was conducted on the monthly trade data using a fixed effects estimator that accounts for cross-country-product heterogeneities. Robust standard errors are reported in parentheses. Asterisk indicate the level of significance, *** p < 0.01, ** p < 0.05, * p < 0.1. Only coefficients for lags are displayed due to space constraint. 41 A3 Sea and air export and import price Table A3: Effects of domestic and external lockdown policies on sea and air export prices Air Sea Lockdown event dummies External Domestic External Domestic Lag0 0.972 -57.534 -0.421 -676.137** (1.633) (150.720) (0.750) (295.726) Lag1 -1.907 35.107 0.338 -625.513** (1.970) (160.169) (0.872) (279.666) Lag2 -1.111 -47.918 -0.646 -606.440** (2.444) (119.554) (1.005) (263.761) Lag3 -3.292 -7.237 1.767 -583.227** (2.945) (123.971) (1.156) (249.142) Lag4 -4.637 -22.341 -2.435** -517.364** (3.351) (121.282) (1.199) (231.761) Lag5 -4.966 59.445 -1.988 -469.588** (3.677) (150.451) (1.305) (215.843) Lag6 -2.217 -11.493 -2.171* -458.945** (3.935) (97.015) (1.296) (200.895) Lag7 -3.281 -34.318 -3.228** -436.885** (4.230) (95.110) (1.347) (185.558) Lag8 -2.713 59.312 -2.103 -375.124** (4.478) (68.370) (1.394) (168.358) Lag9 -0.228 -8.076 -2.519* -319.121** (4.728) (65.198) (1.427) (151.522) Lag10 0.102 -9.062 -4.723*** -317.242** (4.993) (72.658) (1.470) (137.094) Lag11 -2.563 - -5.278*** -274.794** (5.260) - (1.506) (120.185) Lag12 -0.334 - -5.994*** -238.927** (5.566) - (1.533) (104.034) Lag13 -0.951 70.706* -4.544*** -146.949* (5.838) (37.346) (1.544) (84.999) Lag14 -0.767 - -4.448*** -143.740** (6.138) - (1.575) (71.079) Lag15 -1.897 -54.680 -5.033*** -95.291* (6.430) (63.483) (1.608) (56.884) Lag16 2.726 -801.910*** -5.543*** -60.900 (6.783) (64.412) (1.659) (40.380) Lag17 4.097 -786.451*** -1.145 -22.896 (7.139) (72.601) (1.782) (17.091) Lag18 3.325 -172.946* -3.565* - (7.558) (94.528) (1.887) - Lag19 5.523 - -6.696*** 57.084** (8.061) - (2.161) (22.409) Constant 292.219*** -28.502 135.067*** 670.094** (9.492) (292.732) (2.987) (303.843) Observations 449,677 449,677 1,016,792 1,016,792 Number of pair id 70,469 70,469 121,651 121,651 R-squared 0.001 0.001 0.003 0.003 Note: Estimation was conducted on the monthly trade data using a fixed effects estimator that accounts for cross-country-product heterogeneities. Robust standard errors are reported in parentheses. Asterisk indicate the level of significance, *** p < 0.01, ** p < 0.05, * p < 0.1. Only coefficients for lags are displayed due to space constraint. 42 Table A4: Effects of domestic and external lockdown policies on sea and air import prices Air Sea Lockdown event dummies External Domestic External Domestic Lag0 -1.210 1.147 792.943 501.299 (1.009) (0.905) (717.929) (428.716) Lag1 -0.625 3.582*** 822.258 532.894 (1.494) (1.263) (660.556) (413.204) Lag2 -1.243 4.927*** 798.467 419.633 (2.066) (1.656) (600.236) (392.931) Lag3 0.258 8.841*** 686.224 371.908 (2.640) (2.054) (542.455) (368.258) Lag4 2.020 10.588*** 651.027 374.205 (3.195) (2.473) (491.103) (352.693) Lag5 -0.787 12.176*** 647.602 323.544 (3.729) (2.854) (435.285) (334.585) Lag6 -2.703 11.850*** 629.344 286.807 (4.265) (3.243) (384.099) (319.027) Lag7 -5.067 12.136*** 606.955* 249.199 (4.803) (3.622) (340.598) (297.672) Lag8 -6.303 12.159*** 585.874* 271.285 (5.336) (4.002) (315.186) (284.883) Lag9 -7.138 13.004*** 481.776* 242.416 (5.874) (4.384) (282.640) (266.188) Lag10 -7.678 13.230*** 448.605* 200.001 (6.413) (4.770) (260.358) (247.390) Lag11 -12.968* 12.874** 414.934 192.656 (6.956) (5.151) (252.660) (229.207) Lag12 -12.393* 12.106** 1,075.573 157.977 (7.504) (5.537) (875.117) (203.146) Lag13 -13.766* 12.529** 988.707 155.232 (8.057) (5.922) (815.069) (184.049) Lag14 -12.534 13.505** 1,022.361 163.405 (8.613) (6.325) (756.816) (171.345) Lag15 -17.512* 14.120** 847.179 109.410 (9.171) (6.723) (731.808) (144.548) Lag16 -16.885* 13.795* 851.059 123.879 (9.739) (7.136) (721.808) (139.747) Lag17 -18.864* 11.711 813.962 128.091 (10.309) (7.557) (721.003) (121.094) Lag18 -20.917* 13.907* 851.967 44.459 (10.897) (7.992) (698.290) (86.925) Lag19 -21.103* 12.879 - - (11.490) (8.445) - - Constant 329.594*** 202.989*** -467.428 74.555 (7.435) (5.424) (704.517) (132.780) Observations 1,106,040 771,568 1,106,040 771,568 Number of pair id 151,045 110,134 151,045 110,134 R-squared 0.003 0.002 0.003 0.002 Note: Estimation was conducted on the monthly43 trade data using a fixed effects estimator that accounts for cross-country-product heterogeneities. Robust standard errors are reported in parentheses. Asterisk indicate the level of significance, *** p < 0.01, ** p < 0.05, * p < 0.1. Only coefficients for lags are displayed due to space constraint. A4 Import price by exposure to specific NTMs Table A5: Effects of domestic and external lockdown policies on import prices for intermediate products exposed to NTMs Lockdown dummies PSI-External PSI-Domestic PoE-External PoE-Domestic SNI-External SNI-Domestic IA-External IA-Domestic Lag0 -2.807 -3.803** 64.626 13.739 4.234 -3.686 -0.214 -4.240** (2.111) (1.589) (45.009) (62.133) (4.115) (2.296) (2.758) (2.128) Lag1 -7.316*** 6.631*** 80.599 1.964 5.096 7.329*** -5.930* 2.374 (2.675) (1.625) (67.209) (27.207) (6.939) (2.301) (3.254) (2.124) Lag2 -8.353** 10.010*** 154.062* 28.171 11.990 4.688* -3.231 5.195** (3.535) (1.677) (81.484) (40.633) (10.010) (2.437) (3.993) (2.109) Lag3 -4.112 -7.534*** 165.229* -12.938 9.594 2.092 -0.058 -2.941 (4.410) (1.921) (83.317) (35.123) (13.030) (2.610) (4.688) (2.255) Lag4 -3.796 -9.736*** 91.275 5.216 15.780 0.024 5.176 -1.770 (5.184) (2.021) (91.835) (24.466) (16.110) (2.720) (5.347) (2.424) Lag5 -8.679 -10.114*** 123.397 12.641 15.553 0.314 4.286 1.568 (5.887) (1.851) (107.600) (41.457) (19.133) (2.531) (5.916) (2.263) Lag6 -8.863 -7.886*** 103.949 32.644 15.310 0.230 2.036 -2.340 (6.628) (1.792) (146.779) (43.793) (22.167) (2.548) (6.565) (2.248) Lag7 -14.388* -6.090*** 108.792 -4.995 16.258 2.135 -1.289 -0.985 (7.385) (1.768) (181.449) (37.820) (25.224) (2.499) (7.153) (2.214) Lag8 -15.112* -4.101** 216.326 41.560 15.560 1.167 -2.793 2.300 (8.197) (1.803) (223.039) (41.100) (28.317) (2.446) (7.944) (2.213) Lag9 -19.863** -1.287 226.592 1.702 21.122 9.950*** -6.952 7.939*** (9.086) (1.813) (222.176) (31.678) (31.486) (2.563) (8.731) (2.267) Lag10 -20.110** -3.469* 216.573 -8.996 25.190 4.723* -6.804 0.662 (9.999) (1.870) (252.643) (25.597) (34.609) (2.574) (9.539) (2.269) Lag11 -21.595** -7.708*** 214.781 21.576 28.382 5.907** -8.858 2.323 (10.944) (1.800) (250.394) (29.429) (37.851) (2.571) (10.466) (2.232) Constant 294.156*** 316.015*** -37.927 169.969*** 382.841*** 371.295*** 284.158*** 306.768*** (12.030) (1.145) (241.551) (20.541) (40.365) (1.636) (11.827) (1.451) Observations 136,399 136,399 278 278 74,197 74,197 93,938 93,938 Number of pair id 21,711 21,711 42 42 11,911 11,911 15,730 15,730 R-squared 0.006 0.005 0.135 0.050 0.002 0.002 0.002 0.001 Note: Estimation was conducted on the monthly trade data using a fixed effects estimator that accounts for cross-country-product heterogeneities. Robust standard errors are reported in parentheses. Asterisk indicate the level of significance, *** p < 0.01, ** p < 0.05, * p < 0.1. Only coefficients for lags are displayed due to space constraint. 44 A5 Import price by non-exposure to specific NTMs Table A6: Effects of domestic and external lockdown policies on import prices for intermediate products not exposed to NTMs Lockdown event dummies Non PSI-External Non PSI-Domestic Non PoE-External Non PoE-Domestic Non SNI-External Non SNI-Domestic Non IA-External Non IA-Domestic Lag0 1.853 -3.856*** 0.455 -3.706*** -0.128 -3.581*** 0.337 -3.551*** (1.461) (0.915) (1.218) (0.788) (1.289) (0.841) (1.343) (0.856) Lag1 4.952** 5.645*** 1.781 5.935*** 1.120 5.834*** 2.631 6.615*** (2.320) (0.928) (1.881) (0.802) (1.975) (0.860) (2.108) (0.872) Lag2 8.137** 3.693*** 3.895 4.898*** 2.038 5.133*** 4.204 4.824*** (3.283) (0.932) (2.643) (0.814) (2.766) (0.868) (2.980) (0.885) Lag3 11.640*** -2.248** 6.936** -3.050*** 5.373 -3.672*** 6.851* -3.218*** (4.239) (0.987) (3.401) (0.874) (3.554) (0.928) (3.846) (0.952) Lag4 17.246*** 0.667 11.134*** -1.388 9.055** -1.451 10.624** -1.247 (5.207) (1.049) (4.165) (0.931) (4.346) (0.992) (4.718) (1.010) Lag5 18.460*** -0.487 10.786** -2.224** 8.557* -2.443*** 10.123* -2.817*** (6.154) (0.989) (4.907) (0.871) (5.117) (0.931) (5.568) (0.946) Lag6 19.463*** -3.449*** 11.186** -4.364*** 8.515 -4.836*** 10.793* -4.673*** (7.114) (0.992) (5.662) (0.866) (5.901) (0.923) (6.430) (0.941) Lag7 22.803*** -3.996*** 12.748** -4.347*** 10.036 -5.112*** 12.622* -4.950*** (8.083) (1.006) (6.425) (0.875) (6.696) (0.935) (7.304) (0.955) Lag8 22.984** -3.078*** 12.486* -3.073*** 9.489 -3.684*** 12.259 -4.087*** (9.055) (0.981) (7.194) (0.861) (7.495) (0.922) (8.179) (0.936) Lag9 27.579*** 3.238*** 14.972* 2.361*** 11.515 1.334 15.341* 1.453 (10.044) (0.998) (7.980) (0.874) (8.309) (0.932) (9.073) (0.949) Lag10 30.453*** -1.924* 16.795* -2.172** 12.972 -3.085*** 17.313* -2.737*** (11.039) (0.994) (8.770) (0.875) (9.132) (0.932) (9.972) (0.950) Lag11 31.133*** 0.841 16.403* -0.760 11.832 -1.442 16.904 -1.280 (12.063) (0.988) (9.585) (0.864) (9.980) (0.918) (10.898) (0.939) Constant 366.513*** 342.247*** 347.789*** 336.703*** 339.753*** 332.107*** 354.508*** 341.675*** (12.863) (0.648) (10.237) (0.562) (10.664) (0.600) (11.636) (0.613) Observations 512,789 512,789 648,910 648,910 574,991 574,991 555,250 555,250 Number of pair id 63,282 63,282 78,324 78,324 71,750 71,750 69,012 69,012 R-squared 0.001 0.001 0.001 0.001 0.001 0.001 0.002 0.002 Note: Estimation was conducted on the monthly trade data using a fixed effects estimator that accounts for cross-country-product heterogeneities. Robust standard errors are reported in parentheses. Asterisk indicate the level of significance, *** p < 0.01, ** p < 0.05, * p < 0.1. Only coefficients for lags are displayed due to space constraint. 45