## wpiea2021225-print-pdf

## Source details

**Canonical URL:** [wpiea2021225-print-pdf](https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021225-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2021/english/wpiea2021225-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2021/english/wpiea2021225-print-pdf.pdf.json)

---

### Introduction and methodological contributions
- Objective: Provide an application of satellite-based vessel tracking (UN Global Platform) to construct daily indicators of port and trade activity for Pacific islands to fill gaps in official data, provide early warning of turning points, and track shocks in real time.
- Methodological contributions:
  - Most comprehensive methodology for overcoming challenges with vessels’ Automatic Identification System (AIS) data to estimate trade, including:
    - appropriate filtering, backpropagating, historical averaging techniques to overcome possible measurement errors with AIS data;
    - vessel-specific information to avoid estimating trade volumes linearly with draft changes;
    - use of domain expertise, including country/port specific information and liner shipping schedules, to improve the accuracy of estimates.
  - Approach does not require access to commercial data sources.
  - Results and underlying algorithm can be replicated transparently by others with access to the UN Global Platform; the algorithm and defined port boundaries are available from the authors upon request.
- Uses: Useful during normal times for budgeting and policy, and during extreme events (natural disasters, pandemics) for timely responses.

### Stylized facts about Pacific islands — summary findings
- Reliance on imports:
  - Imports of goods make up about 60 percent of GDP for Pacific islands.
  - Exports of goods (excluding re-exports) are less than 10 percent of GDP for all Pacific islands except Fiji (sugar), Nauru (phosphate), and Solomon Islands (timber).
  - Overall, 80–100 percent of all imports (by value) arrive through seaports.
  - Import traffic typically determines vessel traffic at Pacific island ports.
- Maritime connectivity and structure:
  - Pacific islands are among the least connected countries in the world in terms of maritime transport per the Liner Shipping Connectivity Index (LSCI).
  - Trade largely follows a hub-and-spoke model with two major hubs: Guam in the North Pacific and Fiji in the South Pacific; containers arrive at transshipment hubs then move on smaller feeder vessels via "string" services.
  - Example connectivity implication: seven shipping services and 14 port moves would be necessary to export a container from some Pacific islands to Atlantic islands for one trade transaction (UNCTAD, 2020).
- Limited statistical capacity:
  - About half of the Pacific islands are fragile states as defined by the IMF (weak institutional capacity measured by the World Bank CPIA score).
  - GDP data are generally published only on an annual basis, often with a long lag (up to a year or more) and without expenditure composition.
  - More than half of the Pacific island countries publish merchandise trade data with a lag of 12 months or more.
  - There is a positive and usually very high correlation between merchandise imports and GDP for most Pacific islands (Figure 3A).
  - Mirror data from DOTS or bilateral sources may be timelier but often do not match official data closely, possibly because of transshipment and re-export.
- Vulnerability to climate change and natural disasters:
  - Pacific islands rank highest on the risk of suffering a disaster per the World Risk Index; six Pacific islands rank among the first 16 countries at highest risk, the first being Vanuatu.
  - Damages and losses have averaged (median) 9.5 percent of GDP for Pacific islands and over 30 percent of GDP in some cases (examples: Cyclone Evan hitting Samoa in 2012; Cyclone Pam hitting Vanuatu and Tuvalu in 2015; Cyclone Gita hitting Tonga in 2018).
  - Recent event: April 2020, severe tropical Cyclone Harold struck Fiji, Solomon Islands, Tonga, and Vanuatu, causing significant losses and damages including to port infrastructure and jetties.
  - Estimated annual trade disruption due to climate extremes and natural disasters: 1 to 6 percent of the Pacific islands trade (Verschuur et al., forthcoming).
  - Port disruption statistics from vessel tracking analyses:
    - median disruption duration: six days;
    - ninety-fifth percentile disruption duration: 22.2 days;
    - all analyzed events show multiple ports being affected simultaneously.
  - Many Pacific island countries rely heavily on trade/consumption taxes for revenue: almost 62 percent of the Pacific Islands’ tax revenue comes from taxes on goods and services.

### Box 1 — Vulnerability of Pacific Island Ports to Climate Change and Natural Disasters
- Climate exposure and infrastructure needs:
  - Some Pacific islands (Kiribati, Marshall Islands, Tuvalu) have very low elevation (as low as 2 meters on average) and are increasingly exposed to floods and coastal erosion.
  - IMF (2021a) estimates that investment needs for climate-proofing infrastructure will average 3.3 percent of GDP annually for the Asia Pacific region during the next decade, with the amount being much higher for some Pacific island countries.
  - Main ports in the Pacific were built in the 1970s and are reaching the end of their economic lives under the design codes of that time (ADB, 2019), implying large upgrade needs.
- Port function during disasters and design priorities:
  - Ports serve as gateways for international trade and as “lifeline” infrastructure for relief efforts after disasters.
  - Ensuring basic port facilities can withstand disaster impacts so the port remains operational is an important design consideration (ADB, 2020).
  - Upgrading during new investment is relatively inexpensive; retrofitting existing assets is significantly more expensive.
- Current resilience initiatives and project examples:
  - Nauru: ADB is developing a climate-resilient port co-financed by the Green Climate Fund and the Government of Australia.
  - Solomon Islands (Honiara Port): a wharf has been selected as the lifeline wharf and its design checked to lifeline standards.
  - Lifeline-focused design approaches are being considered for Samoa (Apia) and Tonga (Nuku'alofa).
- Implications of global decarbonization of shipping:
  - The IMO adopted an initial GHG emission strategy in 2018 envisaging a reduction in carbon intensity of international shipping by at least 40 percent by 2030.
  - AIS data combined with ship engine and fuel information can help assign CO2 emissions to a vessel’s flag state or territorial waters and could in principle be used to collect levies proportional to emissions.

### Box 2 — Shipping Lines Serving the Pacific Islands and AIS methods
- Major shipping lines and services (examples and characteristics):
  - Kyowa Shipping: three liner services via Guam, Korea (Busan) and Japan; Micronesia Service uses three multi-purpose vessels of 12000 DWT; South Pacific Service uses four multi-purpose vessels of 18000 DWT; Papua New Guinea/Australia Service uses two multi-purpose vessels of 8500 DWT; partnership with Matson.
  - Neptune Pacific Direct Line: formed in 2020; links South Pacific with Australia and New Zealand; owns 50 percent interest in Pacific Forum Line (PFL); partnership with Mariana Express Lines (MELL).
  - Swire Shipping: operates 13 liner services connecting over 400 ports in the Asia-Pacific region; multiple named lines serving Polynesia, Australia Pacific Islands, New Zealand Eastern Pacific Islands, and Pacific North Asia.
- Port identification, boundaries, and coverage:
  - Central port locations identified using WPI and Google Maps; port boundaries (polygons) generated by analyzing vessel movements; analysis covers 30 ports and "all major international ports of Pacific island countries" covering the complete share of their maritime trade (examples listed by country).
- AIS-derived port-call filtering and sample statistics:
  - Filtering: focus on container ships, vehicle carriers, and bulk carriers; exclude fuel tankers and fishing vessels; exclude port calls with a turnaround time of less than 5 hours and no draft change between current and next port.
  - Resulting sample: 3605 port calls by 260 vessels during 2019–20 (Table 1).
- Vessel characteristics, block coefficient, and utilization estimation:
  - AIS provides length, width, draft, and ship type but not design draft or DWT; ship registry databases used to obtain carrying capacity and subtype information.
  - Block coefficients from DHI (2018) and utilization estimation per Jia et al. (2019) are applied (equations and variable definitions preserved in the source).
  - Utilization rate ranges from zero to 100 percent; observed utilization typically between a positive number (e.g., 30–40 percent) and 100 percent due to ballast water.
- Trade flow derivation and corrections:
  - Trade flows derived from differences in utilization entering (μr,in) and leaving (μr,out) a port; change in payload (percentage points) multiplied by vessel DWT yields trade flow in metric tons.
  - Ballast water accounted for implicitly via utilization; when draft is not updated, algorithm backpropagates or uses vessel-specific historical averages (Taverage) for that vessel–port pair.
  - Imports defined as trade flows smaller than zero; exports defined as trade flows larger than zero.
  - To exclude domestic inter-island trade by small vessels, port calls are not included if vessels visited only one country during the time period.
- Hub-and-spoke model and accuracy:
  - Major hubs: Guam (North Pacific) and Fiji (South Pacific); containers moved from hubs to smaller ports via feeder vessels.
  - Identification of liner companies and fleets improved matching with official port call data.
  - Overall, the indicator of trade explains 89   percent of the variation in available official statistics.
- Quality assurance and country validations:
  - Fiji: January 2020 shipping schedule produced a perfect match.
  - Nauru: real-time incoming container ship information (since mid-2020) produced a perfect match.
  - Samoa: monthly "Cargo Shipping Report" comparison produced minor monthly differences but perfect annual match for 2019–20.

### Box 4 — Comparison with Official Trade Data and COVID-19 findings
- Data and caveats:
  - Sample: monthly merchandise trade data for 2019–20 where available; countries with official monthly data from SPC: Kiribati, Palau, Vanuatu for 2019; Fiji, Samoa, Tonga, Tuvalu for 2019–20.
  - Focus excludes fuel, aircraft and boats; data converted into US dollar for comparability.
  - Caveats: official data in value terms vs trade proxy in volume terms; the proxy may lead official statistics by about a month; no monthly trade data in volume terms available for direct comparison.
- Panel regression comparing official data and proxy (sample: Pacific Islands, 2019–20):
  - Trade proxy coefficients:
    - With country fixed effects: 0.79*** (0.15)
    - Without country fixed effects: 1.79*** (0.16)
  - Constant:
    - With country fixed effects: 26.53*** (2.15)
    - Without country fixed effects: 12.36*** (3.73)
  - Observations: 132 (both specifications)
  - Number of countries: 77
  - R-squared:
    - With country fixed effects: 0.89 (within...0.58)
    - Without country fixed effects: 0.89 (between...0.98)
  - Notes: Standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1.
  - Interpretation: Trade proxy and official trade data are reasonably aligned; goodness of fit for pooled data is 0.89 with or without country fixed effects.
- Findings on trade during the COVID-19 pandemic (2019–20):
  - Daily indicators compared January–December 2020 vs January–December 2019 to avoid seasonal effects.
  - Early-stage supply disruptions (January–mid 2020):
    - Vessel traffic and import indicators showed a trade slowdown as early as January/February 2020, before WHO declared COVID-19 a pandemic on March 11, 2020.
    - Port entry restrictions from around March 2020 and lockdowns by major trading partners contributed to supply disruptions; trade already fell below 2019 levels by January/February 2020 for most of the region.
    - Example disruptions: first COVID-19 case in Lautoka, Fiji in March 2020 led to port closure and ship diversions; June 2020 saw some islands in Kiribati experiencing food shortages as ships had not visited since March 2020.
    - Supply impact relatively short-lived, dissipating in the second half of 2020; some islands fully rebounded by end-2020.
  - Later-stage differentiation by tourism dependence (second half of 2020):
    - Tourism-reliant islands experienced larger and more persistent declines in imports compared to islands with lower tourism dependence.
    - International tourism receipts for Fiji, Palau, Samoa, and Vanuatu fell by an estimated 17 to 29 percent of GDP in 2020.
    - Imports dropped as much as 20 percent in some tourism-reliant countries.
    - Official trade value data for 2020 (available for four islands) show Fiji and Samoa (higher tourism reliance) experienced much larger declines in imports than Tonga and Tuvalu.
    - AIS-based trade-volume estimates confirm that import volumes for islands with lower tourism dependence rebounded in the second half of 2020, while they remained low for islands with higher tourism dependence.
- Broader implications and potential applications:
  - Daily near–real-time vessel indicators useful for budget formulation and policy adjustment when official data are delayed.
  - Possible extensions: apply to other small island states (e.g., Caribbean); extend to container-shipping economies with domain expertise; track cruise ships for tourism indicators; monitor vessel day schemes (VDS) for fishing license fees; assess real-time impacts of trade policy; combine with other satellite data (e.g., night lights) for economic activity; monitor regional nodes (e.g., Fiji) to forecast shock propagation.

### ANNEX TABLE I — Central location of Pacific Island ports (selected entries)
- Fiji:
  - Suva — 18°08'00.8"S — 178°25'26.6"E — WPI/Google Maps
  - Lautoka — 17°36'16.3"S — 177°26'18.4"E — WPI/Google Maps
  - Savusavu — 16°46'41.1"S — 179°19'31.6"E — WPI/Google Maps
  - Malau 1/ — 16°21'34.9"S — 179°21'57.5"E — WPI/Google Maps
  - Nabouwalu/Wairiki 1/ — 16°59'37.0"S — 178°41'08.5"E — AIS/Google Maps
- Kiribati:
  - Betio/Tarawa — 1°22'01.9"N — 172°56'07.7"E — WPI/Google Maps
  - Kiritimati — 2°00'24.2"N — 157°29'16.4"W — AIS/Google Maps
- Marshall Islands:
  - Majuro — 7°05'09.0"N — 171°21'54.8"E — WPI/Google Maps
  - Ebeye 2/ — 8°46'44.2"N — 167°44'08.8"E — WPI/Google Maps
- Samoa:
  - Apia — 13°49'34.9"S — 171°45'41.8"W — WPI/Google Maps
- Solomon Islands (examples):
  - Honiara — 9°25'36.8"S — 159°57'36.0"E; 9°24'45.6"S — 160°01'06.2"E — WPI/Google Maps
  - Noro — 8°14'07.4"S — 157°11'47.6"E — WPI/Google Maps
- Tonga:
  - Nuku'alofa — 21°08'13.6"S — 175°11'10.4"W — WPI/Google Maps
  - Neiafu — 18°39'07.3"S — 173°59'06.6"W — WPI/Google Maps
- Tuvalu:
  - Funafuti — 8°30'10.9"S — 179°11'44.0"E — WPI/Google Maps
- Vanuatu:
  - Port Vila — 17°45'22.0"S — 168°18'07.4"E — WPI/Google Maps
  - Santo/Luganville — 15°30'55.1"S — 167°11'18.3"E — WPI/Google Maps
- Notes:
  - WPI = World Port Index.
  - 1/ Specialized ports for exporting specific commodities (e.g., cane sugar, timber).
  - 2/ Ebeye is identified separately from the Kwajalein port that services the U.S. army base in the atoll.

### Implications for monitoring and policy (summary priorities)
- Monitoring:
  - Develop and deploy high-frequency monitoring tools (satellite AIS–based) for real-time surveillance of trade and port activity.
  - Use vessel-based indicators to complement official statistics, detect turning points, and inform timely policy responses.
- Resilience and fiscal implications:
  - Strengthen region-level maritime network resilience, accounting for hub-and-spoke dependencies (notably Guam and Fiji).
  - Prioritize climate adaptation and disaster resilience investments for ports and key network nodes to reduce economic and fiscal vulnerability.
  - Recognize that port disruptions can reduce freight processing capacity, cause delays and depreciation of goods, increase transportation costs via re-routing, and reduce tax revenue affecting public finances.
- Statistical capacity:
  - Strengthen statistical capacity to reduce publication lags for GDP and merchandise trade data where feasible.

*Source: https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021225-print-pdf.pdf*

### REFERENCES .............................................................................................................

### wpiea2021225-print-pdf - REFERENCES .............................................................................................................

### Introduction and methodological contributions
- Objective: Provide an application of satellite-based vessel tracking (UN Global Platform) to construct daily indicators of port and trade activity for Pacific islands to fill gaps in official data, provide early warning of turning points, and track shocks in real time.
- Methodological contributions:
  - Most comprehensive methodology for overcoming challenges with vessels’ Automatic Identification System (AIS) data to estimate trade, including the use of:
    - appropriate filtering, backpropagating, historical averaging techniques to overcome possible measurement errors with AIS data;
    - vessel-specific information to avoid estimating trade volumes linearly with draft changes;
    - use of domain expertise, including country/port specific information and liner shipping schedules, to improve the accuracy of estimates.
  - Approach does not require access to commercial data sources.
  - Results and underlying algorithm can be replicated transparently by others with access to the UN Global Platform; the algorithm and defined port boundaries are available from the authors upon request.
- Uses: Useful during normal times for budgeting and policy, and during extreme events (natural disasters, pandemics) for timely responses.

### Stylized facts about Pacific islands — summary findings
- Reliance on imports:
  - Imports of goods make up about 60 percent of GDP for Pacific islands.
  - Exports of goods (excluding re-exports) are less than 10 percent of GDP for all Pacific islands except Fiji (sugar), Nauru (phosphate), and Solomon Islands (timber).
  - Overall, 80–100 percent of all imports (by value) arrive through seaports.
  - Import traffic typically determines vessel traffic at Pacific island ports.
- Maritime connectivity and structure:
  - Pacific islands are among the least connected countries in the world in terms of maritime transport per the Liner Shipping Connectivity Index (LSCI).
  - Trade largely follows a hub-and-spoke model with two major hubs: Guam in the North Pacific and Fiji in the South Pacific; containers arrive at transshipment hubs then move on smaller feeder vessels via "string" services.
  - Example connectivity implication: seven shipping services and 14 port moves would be necessary to export a container from some Pacific islands to Atlantic islands for one trade transaction (UNCTAD, 2020).
- Limited statistical capacity:
  - About half of the Pacific islands are fragile states as defined by the IMF (weak institutional capacity measured by the World Bank CPIA score).
  - GDP data are generally published only on an annual basis, often with a long lag (up to a year or more) and without expenditure composition.
  - More than half of the Pacific island countries publish merchandise trade data with a lag of 12 months or more.
  - There is a positive and usually very high correlation between merchandise imports and GDP for most Pacific islands (Figure 3A).
  - Mirror data from DOTS or bilateral sources may be timelier but often do not match official data closely, possibly because of transshipment and re-export.
- Vulnerability to climate change and natural disasters:
  - Pacific islands rank highest on the risk of suffering a disaster per the World Risk Index; six Pacific islands rank among the first 16 countries at highest risk, the first being Vanuatu.
  - Damages and losses have averaged (median) 9.5 percent of GDP for Pacific islands and over 30 percent of GDP in some cases (examples: Cyclone Evan hitting Samoa in 2012; Cyclone Pam hitting Vanuatu and Tuvalu in 2015; Cyclone Gita hitting Tonga in 2018).
  - Recent event: April 2020, severe tropical Cyclone Harold struck Fiji, Solomon Islands, Tonga, and Vanuatu, causing significant losses and damages including to port infrastructure and jetties.
  - Estimated annual trade disruption due to climate extremes and natural disasters: 1 to 6 percent of the Pacific islands trade (Verschuur et al., forthcoming).
  - Port disruption statistics from vessel tracking analyses:
    - median disruption duration: six days;
    - ninety-fifth percentile disruption duration: 22.2 days;
    - all analyzed events show multiple ports being affected simultaneously.
  - Many Pacific island countries rely heavily on trade/consumption taxes for revenue: almost 62 percent of the Pacific Islands’ tax revenue comes from taxes on goods and services.

### Implications for monitoring and policy
- High-frequency vessel-based indicators:
  - Can complement official statistics on merchandise trade and help track economic activity in data-scarce environments.
  - Useful for detecting turning points, monitoring shocks in real time, and informing timely policy responses.
- Network resilience:
  - Importance of resilience at the regional maritime network level, not only at individual country ports; climate-proofing investments should consider hub-and-spoke dependencies (Guam and Fiji as major hubs).
- Policy-relevant risks:
  - Port disruptions can reduce freight processing capacity, cause delays, depreciation of goods, and additional transportation costs via re-routing.
  - Disruptions to trade flows can reduce tax revenue and affect public finances and the government’s ability to maintain public services.
- Suggested priorities:
  - Develop and deploy high-frequency monitoring tools (satellite AIS–based) for real-time surveillance of trade and port activity.
  - Strengthen statistical capacity to reduce publication lags for GDP and merchandise trade data where feasible.
  - Prioritize climate adaptation and disaster resilience investments for ports and key nodes in the maritime network to reduce economic and fiscal vulnerability.

*Source: https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021225-print-pdf.pdf*

### Box 1. Vulnerability of Pacific Island Ports to Climate Change and Natural Disasters

### Box 1. Vulnerability of Pacific Island Ports to Climate Change and Natural Disasters

### Climate exposure and infrastructure needs
- The Pacific islands are situated in one of the most disaster-prone geographical areas; they face climate change-related extreme weather events, such as Cyclone Harold in April 2020.
- Some Pacific islands, such as Kiribati, Marshall Islands, and Tuvalu, have very low elevation (as low as 2 meters on average) and are increasingly exposed to floods and coastal erosion.
- IMF (2021a) estimates that investment needs for climate-proofing infrastructure will average 3.3 percent of GDP annually for the Asia Pacific region during the next decade, with the amount being much higher for some Pacific island countries.
- Main ports in the Pacific were built in the 1970s and are reaching the end of their economic lives under the design codes of that time (ADB, 2019), implying large upgrade needs to address risks from cyclones, floods, storm surges, and sea level rise.

### Port function during disasters and design priorities
- Ports serve both as gateways for international trade and as “lifeline” infrastructure to support relief efforts after disasters.
- An important design consideration is ensuring basic port facilities can withstand disaster impacts so the port remains operational in the aftermath (ADB, 2020).
- Upgrading can be relatively inexpensive if it involves making new investment projects climate resilient; retrofitting existing climate-exposed critical assets or developing coastal protection infrastructure is significantly more expensive.

### Current resilience initiatives and project examples
- Port renewal and expansion plans in the Pacific increasingly adopt climate-resilient approaches (ADB, 2019 and 2020).
- Nauru: ADB is developing a climate-resilient port that can operate year-round; the new port is co-financed by the Green Climate Fund and the Government of Australia.
- Solomon Islands (Honiara Port): A wharf has been selected as the lifeline wharf and its design checked to the more rigorous standards required for a lifeline structure.
- Similar lifeline-focused design approaches are being considered for new port facilities in Samoa (Apia) and Tonga (Nuku'alofa).

### Implications of global decarbonization of shipping
- In 2018, the IMO adopted its initial greenhouse gas (GHG) emission strategy that envisages a reduction in carbon intensity of international shipping by at least 40 percent by 2030.
- Initiatives such as the Getting to Zero Coalition, supported by UNCTAD, also aim at reducing CO2 emissions from shipping to net-zero.
- A ship’s emissions depend on numerous factors, including the fuel used and speed.
- AIS data—in combination with information about the ship’s engine and fuel—can help assign CO2 emissions to the country of the vessel’s flag or the country’s waters where the CO2 is being emitted; this could in principle be used to collect levies proportional to the emissions by the flag state or the state where the emissions take place (Parry et al. 2018).

*Source: Box 1. Vulnerability of Pacific Island Ports to Climate Change and Natural Disasters (wpiea2021225-print-pdf).*

### Box 2. Shipping Lines Serving the Pacific Islands

### Box 2. Shipping Lines Serving the Pacific Islands

### Major shipping lines and services
- Kyowa Shipping
  - Provides three liner services via Guam, Korea (Busan) and Japan connecting European and Asian ports to the Pacific Islands.
  - Micronesia Service: serves Micronesian ports, mainly using three multi-purpose vessels of 12000 DWT. Ports of call: Palau (Koror); Federated States of Micronesia (Yap, Chuuk, Pohnpei, Kosrae); Marshall Islands (Majuro, Kwajalein, Ebeye). Ports visits take place about every two weeks.
  - South Pacific Service: provides service to the South Pacific with four multi-purpose vessels of DWT 18000 DWT. Ports of call: Kiribati (Tarawa); Solomon Islands (Honiara); Vanuatu (Port Vila, Santo); New Caledonia (Noumea); Fiji (Suva, Lautoka); Tonga (Nuku'alofa); Samoa (Apia); American Samoa (Pago Pago); Tahiti (Papeete).
  - Papua New Guinea/Australia Service: provides service with two multi-purpose vessels of 8500 DWT. Ports of call: Papua New Guinea (Lae, Port Moresby, Rabaul); Australia (Townsville).
  - Partnership with the US-based Matson shipping company.
- Neptune Pacific Direct Line
  - Formed in 2020, when Neptune Pacific Line acquired Pacific Direct Line (PDL).
  - Links the South Pacific Islands with Australia and New Zealand.
  - Has a cooperation with the government of Samoa and owns a 50 percent controlling interest in Pacific Forum Line (PFL).
  - Vessels call in Vanuatu (Vila and Santo), Fiji (Suva and Lautoka), Tonga (Nuku'alofa), Samoa (Apia), Tahiti (Papeete), Tuvalu (Funafuti), Solomon Islands (Honiara), as well as Kiribati, Marshall Islands, and Nauru.
  - Partnership with Mariana Express Lines (MELL) to service Palau and Yap out of Guam.
- Swire Shipping
  - Operates 13 liner services connecting over 400 ports in the Asia-Pacific region.
  - Polynesia Line: shipping network from the U.S. West Coast to the Pacific Islands, offering a fortnightly service from Los Angeles and Oakland in the United States to Tahiti, Tonga, Samoa, American Samoa, and the wider Pacific.
  - Australia Pacific Islands line: multi-purpose liner service accepting containerized, breakbulk and project cargo between East Australia and New Caledonia, Vanuatu, Fiji, Samoa, and Tonga.
  - New Zealand Eastern Pacific Islands line: service between Auckland and Tonga, Fiji, Samoa, Tahiti, the Cook Islands, and Niue.
  - Pacific North Asia line: services between North Asia and the Marshall Islands, Kiribati, Vanuatu, Solomon Islands, New Caledonia, Fiji, Tonga, Samoa, and Tahiti.

### Port identification, boundaries, and coverage
- Central port locations are identified using WPI and Google Maps.
- Port boundaries (polygons) are generated by analyzing vessel movements in and around the port using satellite imagery.
- Examples provided: port boundaries for Nauru and Samoa (Figure 5A referenced).
- The analysis covers 30 ports, described as all major international ports of Pacific island countries and covering the complete share of their maritime trade. Examples include:
  - Fiji: Suva, Lautoka, Malau, Nabouwalu, Savusavu
  - Kiribati: Betio/Tarawa, Kiritimati
  - Marshall Islands: Majuro, Ebeye
  - Federated States of Micronesia: Yap, Chuuk, Pohnpei, Kosrae
  - Nauru: Nauru port
  - Palau: Malakal port
  - Samoa: Apia port
  - Solomon Islands: Honiara, Noro, Mbulo, Ndora, Gizo, Rennell Island, Lughughi, Ringgi, Seghe
  - Tonga: Nuku'alofa, Neiafu
  - Tuvalu: Funafuti
  - Vanuatu: Port Vila, Santo/Luganville

### AIS-derived port-call filtering and sample statistics
- Filtering criteria to extract port calls that contribute to trade:
  - Focus on container ships, vehicle carriers, and bulk carriers. Fuel tankers are excluded. Fishing vessels are excluded.
  - Exclude port calls with a turnaround time of less than 5 hours and no draft change between the current and next port (to filter out vessels in transit).
- Resulting sample: 3605 port calls by 260 vessels during 2019–20 (Table 1).
- The algorithm and defined port boundaries are available from the authors upon request.
- An alternative experimented approach used a pre-determined port boundary (a circle around central port locations); it generated similar results given the relatively simple structure of Pacific island ports.

### Vessel characteristics, block coefficient, and utilization estimation
- AIS data provides vessel length, width, draft, and ship type but does not include maximum (design) draft, carrying capacity (DWT), or ship subtype details.
- Ship registry databases from FleetMon and IHS Markit are used to obtain carrying capacity and subtype information.
- Block coefficients for vessel types and multiple DWT indicators per type are obtained from DHI (2018) and matched by closest DWT per vessel type.
- Utilization rate estimation follows Jia et al. (2019). Notation and formulae as presented:
  - Equation (1)
    휇휇
    푟푟
    =
    (
    퐶퐶푏푏
    푟푟
    푑푑
    푟푟
    −퐶퐶푏푏
    푑푑
    푑푑
    푑푑
    )
    퐿퐿퐿퐿ρ
    푤푤
    +퐷퐷퐿퐿퐷퐷
    퐷퐷
    퐿퐿
    퐷퐷
  - Variable definitions provided verbatim:
    - L=length; W=width; d
    d
    =design (maximum) draft; d
    r
    =reported draft; Cb
    d
    =block coefficient at design draft; Cb
    r
    =block coefficient at reported draft; ρ
    w
    = average density of salt water (1.025 tons/m3); DWT=deadweight tonnage of vessel (i.e. maximum carrying capacity of the vessel when it is floating at its maximum draft); and μ
    r
    = vessel payload at reported draft.
  - Equation (2)
    퐶퐶푏푏
    푟푟
    = 1−
    �
    (
    1−퐶퐶푏푏
    푑푑
    )
    푑푑
    푑푑
    푑푑
    푟푟
    1/3
    �
  - Backpropagation of outgoing draft when not updated (Equation (3)):
    푑푑
    푟푟,표표표표표표
    =
    �
    푑푑
    푟푟,푖푖푖푖,푖푖푛푛   푛푛표표

    푑푑
    푟푟,표표표표표표
        푖푖푖푖    푑푑
    푟푟,푖푖푖푖
    =푑푑
    푟푟,표표표표표표
         푖푖푖푖     푑푑
    푟푟,푖푖푖푖
    ≠푑푑
    푟푟,표표표표표표
  - Explanation: utilization rate ranges from zero to 100 percent; observed utilization typically between a positive number (e.g., 30–40 percent) and 100 percent due to ballast water present in ships.
- Block coefficients and vessel characteristics are used to estimate payload (utilization rate) when entering and leaving the port boundary.

### Trade flow derivation and corrections
- Trade flows derived from differences in utilization when entering (μ
  r,in
  ) and leaving a port (μ
  r,out
  ). The change in payload (percentage points) multiplied by vessel DWT yields trade flow in metric tons.
- Ballast water is accounted for implicitly through utilization; no separate ballast water assumption required as long as ballast water does not change within the port boundary.
- When no draft change is observed due to non-updated dynamic information, the algorithm estimates the average trade flow associated with a specific vessel in a specific port based on the vessel’s history of port calls.
- Estimated trade flow expression (Equation (4)):
  퐷퐷=�
  �
  휇휇
  푟푟,표표표표표표
  −휇휇
  푟푟,푖푖푖푖
  � 퐷퐷퐿퐿퐷퐷
  퐷퐷
  푎푎푎푎푛푛푟푟 푎푎푎푎푛푛
      푖푖푖푖    푑푑
  푟푟,푖푖푖푖
  ≠푑푑
  푟푟,표표표표표표
       푖푖푖푖     푑푑
  푟푟,푖푖푖푖
  =푑푑
  푟푟,표표표표표표
- T
  average
  denotes the average of all non-zero trade flows observed for a given vessel at a given port and is assigned only to vessels that are not in transit.
- Imports are defined as trade flows smaller than zero; exports are defined as trade flows larger than zero.
- To exclude domestic inter-island trade conducted by small vessels, the algorithm does not include port calls if vessels have only visited one country during the time period.
- Domain expertise and liner shipping schedules are used to check direction of trade, especially for specialized export ports and vessel types (e.g., bulk carrier).

### Hub-and-spoke model and accuracy
- Pacific islands’ most important trade flows follow a hub-and-spoke model:
  - Major hubs: Guam in the North Pacific and Fiji in the South Pacific.
  - Large container ships arrive at hubs; containers are moved onto smaller feeder vessels serving smaller and more distant ports.
- Identification of key liner shipping companies and vessel fleets serving the region improved accuracy of identified port calls and matching with official port call data where available.
- Overall, the indicator of trade explains 89   percent of the variation in available official statistics.

### Quality assurance and country validations
- Validation comparisons with official statistics:
  - Fiji: publicly available shipping schedule information (January 2020) used for validation; generated a perfect match.
  - Nauru: real-time incoming container ship information (since mid-2020) used for validation; generated a perfect match.
  - Samoa: detailed monthly "Cargo Shipping Report" (available from 2015 onward) compared with AIS-derived data; with minor monthly differences, annual figures for 2019–20 matched perfectly.

*Source: Box 2. Shipping Lines Serving the Pacific Islands (from the provided PDF content).*

### Box 4. Comparison with Official Trade Data

### Box 4. Comparison with Official Trade Data

### Data and caveats
- Sample: monthly merchandise trade data for the period 2019–20 for Pacific islands where available.
- Countries with official monthly data from SPC (customs data): Kiribati, Palau, and Vanuatu for 2019; Fiji, Samoa, Tonga, and Tuvalu for 2019–20.
- Focus: merchandise imports of goods (excluding fuel, aircraft and boats). Data converted into US dollar for cross-comparability.
- Key caveats:
  - Official data are in value terms, while the trade proxy is in volume terms by construction; a perfect match is not expected.
  - The trade proxy may lead official trade statistics (e.g., by a month) because national customs recording may occur after unloading at port.
  - No monthly trade data for Pacific islands on volume terms are available for direct comparison, highlighting a data gap.

### Panel regression comparing official data and proxy
- Regression specification: official trade data x_it regressed on trade proxy x*_it (panel regression with and without country fixed effects).
- Sample: Pacific Islands, 2019–20.
- Box Table 1 (panel regression results):
  - Columns: (1) With country fixed effects; (2) Without country fixed effects.
  - Trade proxy coefficients:
    - With country fixed effects: 0.79*** (0.15)
    - Without country fixed effects: 1.79*** (0.16)
  - Constant:
    - With country fixed effects: 26.53*** (2.15)
    - Without country fixed effects: 12.36*** (3.73)
  - Observations: 132 (both specifications)
  - Number of countries: 77
  - R-squared:
    - With country fixed effects: 0.89 (within...0.58)
    - Without country fixed effects: 0.89 (between...0.98)
  - Notes: Standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1.
- Interpretation: The trade proxy and official trade data are reasonably aligned; the goodness of fit (R-squared) for the panel regression pulling all available data together is 0.89 with or without country fixed effects.

### Findings on trade during the COVID-19 pandemic (2019–20)
- Timing and methodology:
  - Daily indicators of port and trade activity constructed from satellite-based vessel tracking (UN Global Platform) were compared year-over-year: January–December 2020 vs January–December 2019 to avoid seasonal fluctuations.
- Early-stage supply disruptions (January–mid 2020):
  - Vessel traffic and import indicators highlighted a trade slowdown as early as January/February 2020, before WHO declared COVID-19 a pandemic on March 11, 2020.
  - Port entry restrictions by Pacific islands starting around March 2020 and lockdowns by major trading partners (notably China) contributed to supply disruptions.
  - Trade was already falling below 2019 levels by January/February 2020 for most of the region.
  - Examples of disruption and shortages:
    - When the first case of COVID-19 was recorded in Lautoka, Fiji, in March 2020, the port was closed and ships were diverted to Suva or elsewhere.
    - In June 2020, certain islands in Kiribati began to experience shortages of food products, as ships had not visited some islands since March 2020.
  - The supply impact appears relatively short-lived, dissipating in the second half of 2020; for some islands there was a full rebound in imports by end-2020.
- Later-stage differentiation by tourism dependence (second half of 2020):
  - Tourism-reliant islands experienced larger and more persistent declines in imports compared to islands with lower tourism dependence.
  - Tourism accounts for a substantial share of GDP and employment in several Pacific islands, particularly Fiji, Palau, Samoa, and Vanuatu.
  - International tourism receipts for Fiji, Palau, Samoa, and Vanuatu fell by an estimated 17 to 29 percent of GDP in 2020.
  - With tourism revenues collapsing after Q1 2020, imports dropped as much as 20 percent in some tourism-reliant countries.
  - Official trade value data for 2020 (available for four islands) show:
    - Fiji and Samoa (higher reliance on tourism) experienced much larger declines in imports compared to Tonga and Tuvalu (lower or no dependence on tourism).
  - Trade-volume estimates based on AIS vessel traffic confirm the differentiation: import volumes for islands with lower tourism dependence rebounded in the second half of 2020, while they remained low for islands with higher tourism dependence.

### Broader implications and uses of the methodology
- The daily, near–real-time indicators based on vessel tracking can be useful when timely official data are unavailable, e.g., for budget formulation or policy adjustment during crises.
- The approach highlights the importance of maritime transport as an essential sector for delivery of critical supplies and global trade during crises.
- Potential extensions and applications (methodology can be adapted and extended):
  - Small island states: extend globally to other small island states (e.g., Caribbean).
  - Global supply chains: extend to economies active in container shipping, noting the need for domain expertise on liner shipping operations.
  - Tourism: with modifications, track tourism via cruise ships as a high-frequency indicator for cruise-reliant countries.
  - Fishing license fees: track vessel day schemes (VDS) by estimating days fishing vessels spend at sea to monitor license-fee receipts.
  - Trade policy: assess real-time impact of trade policies (tariffs, sanctions, liberalizations) on trade patterns.
  - Early indicator for economic activity: use for countries relying on maritime trade; combine with other satellite data (e.g., night lights).
  - Regional and network-level analysis: monitor critical nodes (e.g., transshipment hubs like Fiji) to forecast trade shocks and propagation across the network given transit times and limited rerouting options in a rigid network.

*Source: Authors’ estimates (Box 4. Comparison with Official Trade Data).*

### ANNEX TABLE I. CENTRAL LOCATION OF PACIFIC ISLAND PORTS

### ANNEX TABLE I. CENTRAL LOCATION OF PACIFIC ISLAND PORTS

### Ports and coordinates
- Fiji
  - Suva — 18°08'00.8"S — 178°25'26.6"E — WPI/Google Maps
  - Lautoka — 17°36'16.3"S — 177°26'18.4"E — WPI/Google Maps
  - Savusavu — 16°46'41.1"S — 179°19'31.6"E — WPI/Google Maps
  - Malau 1/ — 16°21'34.9"S — 179°21'57.5"E — WPI/Google Maps
  - Nabouwalu/Wairiki 1/ — 16°59'37.0"S — 178°41'08.5"E — AIS/Google Maps
- Kiribati
  - Betio/Tarawa — 1°22'01.9"N — 172°56'07.7"E — WPI/Google Maps
  - Kiritimati — 2°00'24.2"N — 157°29'16.4"W — AIS/Google Maps
- Marshall Islands
  - Majuro — 7°05'09.0"N — 171°21'54.8"E — WPI/Google Maps
  - Ebeye 2/ — 8°46'44.2"N — 167°44'08.8"E — WPI/Google Maps
- Federated States of Micronesia (FSM)
  - Yap — 9°30'56.5"N — 138°07'31.4"E — WPI/Google Maps
  - Chuuk — 7°26'44.5"N — 151°50'23.0"E — WPI/Google Maps
  - Pohnpei — 6°58'46.6"N — 158°12'03.9"E — WPI/Google Maps
  - Kosrae — 5°21'01.7"N — 162°57'24.0"E — WPI/Google Maps
- Nauru
  - Nauru — 0°31'46.9"S — 166°54'26.9"E — WPI/Google Maps
- Palau
  - Malakal — 7°19'51.1"N — 134°27'25.2"E — WPI/Google Maps
- Samoa
  - Apia — 13°49'34.9"S — 171°45'41.8"W — WPI/Google Maps
- Solomon Islands
  - Honiara — 9°25'36.8"S — 159°57'36.0"E; 9°24'45.6"S — 160°01'06.2"E — WPI/Google Maps
  - Noro — 8°14'07.4"S — 157°11'47.6"E — WPI/Google Maps
  - Mbulo — 9°35'24.0"S — 160°37'01.2"E — AIS/Google Maps
  - Ndora 1/ — 8°23'27.6"S — 157°33'14.4"E — AIS/Google Maps
  - Seghe 1/ — 8°34'59.8"S — 157°56'01.6"E — AIS/Google Maps
  - Gizo 1/ — 8°5'60.0"S — 156°50'60.0"E — WPI/Google Maps
  - Ringgi 1/ — 8°7'00.0"S — 157°06'00.0"E — WPI/Google Maps
  - Rennell Island 1/ — 11°41'06.0"S — 160°18'06.5"E — AIS/Google Maps
  - Lughughi 1/ — 11°37'42.6"S — 160°05'16.8"E — AIS/Google Maps
- Tonga
  - Nuku'alofa — 21°08'13.6"S — 175°11'10.4"W — WPI/Google Maps
  - Neiafu — 18°39'07.3"S — 173°59'06.6"W — WPI/Google Maps
- Tuvalu
  - Funafuti — 8°30'10.9"S — 179°11'44.0"E — WPI/Google Maps
- Vanuatu
  - Port Vila — 17°45'22.0"S — 168°18'07.4"E — WPI/Google Maps
  - Santo/Luganville — 15°30'55.1"S — 167°11'18.3"E — WPI/Google Maps

### Notes
- WPI = World Port Index.
- 1/ Specialized ports for exporting specific commodities (e.g., cane sugar, timber).
- 2/ Ebeye is identified separately from the Kwajalein port that services the U.S. army base in the atoll.

*Source: ANNEX TABLE I. CENTRAL LOCATION OF PACIFIC ISLAND PORTS (wpiea2021225-print-pdf).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021225-print-pdf.pdf_