## 1. A Automatic Identification System Messages

## Source details

**Canonical URL:** [1. A Automatic Identification System Messages](https://www.imf.org/-/media/files/publications/cr/2023/english/1slbea2023002.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2023/english/1slbea2023002.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2023/english/1slbea2023002.pdf.json)

---

### A. Introduction
- Purpose and scope:
  - Implements a simplified version of Arslanalp, Koepke and Verschuur (2021) using satellite-based Automatic Identification System (AIS) data to nowcast monthly trade flows and estimate economic disruptions (case study: Solomon Islands).
  - Motivations: Pacific Islands Countries (PICs) typically have capacity constraints that prevent timely high-frequency data; about 80 percent of Solomon Islands trade is carried by sea and shipments in each port tend to be in one direction (focusing either in exports or imports).
- Outputs:
  - Monthly estimators of port activity (total, by port, by activity, inter-island).
  - Export and import volume estimates using AIS data and AKV-based simplified methods.

### B. Data and Methodology
- AIS system and coverage:
  - AIS broadcasts static and voyage-related information every 6 minutes and dynamic information every 2-10 seconds when moving (every 6 minutes when anchored).
  - AIS is compulsory for international commercial ships with gross tonnage (GT) of 300 or more tons, and all passenger ships regardless of size.
  - AIS data accessed via the UN Global Platform (UNGP) with global archive since December 1, 2018.
  - AIS data pulled monthly for January 2019 to January 2023.
- Message types and frequencies:
  - Static, Identity related: Ship callsign, IMO number, mmsi number, vessel name, vessel type, size, flag. Source: Manual. Frequency: Every 6 minutes.
  - Static, Voyage-related: Destination, Estimated Time of Arrival (ETA), draught. Source: Manual input by crew. Frequency: Every 6 minutes.
  - Dynamic: Vessel speed, geospatial location (latitude and longitude), heading, time. Source: Automatic. Frequency: 2-10 seconds when moving, every 6 minutes when anchored.
- Spatial processing and port-call definitions:
  - AIS data stored using Uber’s hexagons definition; port areas use AKV port boundaries covered with level 8 hexagons.
  - Port call defined as a vessel’s visit with clearly defined entry and exit; observations within port boundaries are pulled monthly.
  - Algorithmic processing: combines port calls crossing month boundaries and drops vessels that appear only once within port boundaries.
- Indicators compiled:
  - Port activity indicators (use all port calls): overall port activity, inter-island traffic, port activity by port, port activity by economic activity.
  - Economic-activity categories: law enforcement (military and law enforcement vessels), fuel (tanker vessels), fishing, tourism (passenger, diving, pleasure craft, sailing), trade (bulk carrier, container ships, forest-product carrier, towing, tug, and general cargo vessels). Remaining ships form a residual port activity indicator.
  - Inter-island traffic: counts vessels observed in more than one Solomon Islands’ port within the same period.
- Trade indicators and volume estimation (AKV-based, simplified):
  - Consider only trade-related port calls: vessel type Cargo (bulk carrier, container ships, forest-product carrier, and general cargo vessels).
  - Discard port calls shorter than 5 hours with no change in adjusted draught.
  - Estimate trade volumes from change in utilization rate when entering and leaving port boundary.
  - Utilization rate estimation uses vessel characteristics (length L, width W, carrying capacity DWT, block coefficient) and draft information; average density of salt water used is 1.025 tons/m3.
- Imputation strategies for missing draft information:
  - Local backpropagation: uses incoming draft in the following port when the next port visited by a ship is within Solomon Islands.
  - Historical averaging: imputes draft changes with historical average observed in the port for the ship with missing data.
- Conventions and interpretation:
  - Imports correspond to estimates below zero; exports correspond to estimates above zero.
  - Context: goods imports represent 30 percent of GDP (2017-2021 average). 80 percent of merchandise imports take place by sea. Annual imports have a correlation coefficient of 0.94 with GDP (measured in US dollars, 1995 to 2021).

### C. Port Activity — Findings and Patterns
- Aggregate and temporal patterns:
  - Honiara port heat maps show clear decline in activity during February 2021 and 2022 compared to February 2020.
  - Total port activity decreased by 27 percent in 2020 and by 42 percent in 2021 relative to the 2019 average.
  - Port activity in 2022 remained subdued at the 2021 depressed level but displayed an increasing trend and signs of systematic recovery.
  - Inter-island traffic broadly follows the same pattern as total port activity.
- Port-level and sectoral patterns:
  - Activity concentrated in Honiara (main imports port) and Port Noro (main exports port); other ports considerably less active on average.
  - Port activity declined across all ports with the onset of the pandemic, bottoming out in early 2022 with the domestic outbreak of COVID-19.
  - Recovery trend observed for Honiara, Port Noro, Seghe port, and Gizo Harbor; other ports remain depressed, indicating geographically unequal recovery.
- Economic-activity breakdown:
  - Port calls related to fishing, tourism and trade plummeted with the pandemic; fuel-related port calls fell but remained relatively stable.
  - Incipient recovery in 2022 appears broad-based across fishing, tourism, and trade.
  - Recovery in fishing activity can proxy for recovery of fishing rents in the Balance of Payments; tourism recovery follows border reopening in mid-2022.
- High-frequency event analysis — November 2021 Honiara riots:
  - Heat maps for the week before, during, and after the November 2021 riots show increased water activity in the week of the riots and the week after.
  - Weekly statistics and interpretations:
    - Port calls fell by 30 percent in the week of the riots (week 0) relative to the week before (week -1).
    - Trade-related port calls fell after the riots.
    - Law enforcement port calls increased during the riots period.
    - Ships that arrived the week before the riots spent on average more time in port than those arriving in the week of the riots and the week after, supporting interpretation that ships faced difficulty completing tasks on land.
  - Conclusion: while port calls declined, increased in-water activity suggests ships faced difficulty unloading and there was increased law enforcement presence.

### D. Trade Flow Estimates — Findings and Validation
- Estimation strategies and preferred choices:
  - Two estimation scopes: using information from all ports and using only Honiara port.
  - Validation:
    - Exports validated against monthly volume of logs production and Monthly Production Index (CBSI), and compared with AKV estimates.
    - Imports validation limited; compared only with AKV volume estimates.
  - Preferred estimates:
    - Exports: computed using information from all ports.
    - Imports: computed using information from the Honiara port.
- Pairwise correlations (January 2019 to January 2023; AKV variables sample January 2019 to February 2021):
  - Exports all Ports vs Logs (Volume): 0.44
  - Exports all Ports vs Monthly Production Index: 0.36
  - Exports all Ports vs Exports AKV: 0.87
  - Exports Honiara vs Logs (Volume): -0.49
  - Exports Honiara vs Monthly Production Index: -0.52
  - Exports Honiara vs Exports AKV: 0.12
  - Imports all Ports vs Imports AKV: -0.03
  - Imports Honiara vs Imports AKV: 0.44
  - Exports AKV vs Logs (Volume): 0.56
  - Exports AKV vs Monthly Production Index: 0.46
- Key quantitative findings on trade dynamics:
  - Imports:
    - Import volumes show high dynamism, responding quickly to external and internal conditions.
    - Differences between the import volume estimate and the value indicator suggest increases in shipping costs and global supply chain disruptions (second half of 2021) and increases in shipping costs and global commodity prices in 2022 due to the war in Ukraine.
  - Exports:
    - Exports estimated to have fallen by 64 percent in 2021 and 2022 with respect to 2019.
    - Export estimates show a large boom at end-2019 and beginning-2020 that requires validation with other data sources and authorities.
- Visualization notes:
  - Export and import volume estimates plotted against CBSI monthly merchandise value series (typically lagged by two to three months); figures use six-month moving averages for smoothing.

### E. Conclusions, Implications, and IMF Capacity-Building
- Core conclusions:
  - AIS-based vessel-tracking indicators provide timely, granular proxies for port activity and trade flows that can fill high-frequency data gaps in PICs.
  - For Solomon Islands:
    - AIS indicators captured substantial pandemic-related declines in port activity and heterogeneous recovery across ports and activities.
    - High-frequency analysis identified disruptions around the November 2021 Honiara riots (30 percent weekly drop in port calls; increased law enforcement presence; evidence of ships unable to unload).
    - Trade-flow estimates indicate persistent export weakness (64 percent decline in 2021 and 2022 vs 2019) and volatile import dynamics influenced by shipping costs and global commodity prices.
  - Preferred estimation choices informed by correlation analysis: use all ports for exports and Honiara port for imports.
- Use for surveillance and technical assistance:
  - AIS-derived estimators have been used in the Solomon Islands surveillance dashboard and discussed with authorities.
  - CBSI staff is exploring incorporation of these indicators into broader nowcasting efforts under IMF technical assistance (TA).
  - The IMF can serve as a capacity-building and innovation hub by integrating AIS-derived estimators into surveillance dashboards, providing TA to central banks to incorporate AIS indicators into nowcasting frameworks, and leveraging cross-country synergies and technical expertise to produce AKV estimations.
- Potential applications:
  - Single-country: monitoring fishing vessels to estimate fishing rents; monitoring export-related ships for piracy/exports misreporting; tracking tourism in real time.
  - Global: analyzing global supply chains; assessing trade disruptions from natural disasters; evaluating effects of trade policies.
- Visualization conventions used in figures:
  - Black bars: monthly port calls of all types.
  - Blue line: six-month moving average.
  - Horizontal lines: yearly averages.
  - Vertical red lines: start of the COVID-19 pandemic (March 2020), the November 2021 Honiara riots, and the domestic outbreak of COVID-19 (February 2022).

*Prepared by Gabriela Cugat (RES); IMF staff estimates based on AIS data accessed via the UN Global Platform and AKV methodologies. Source: 1slbea2023002.*

### 1. A

### 1. A Automatic Identification System Messages

### A. Introduction
- Pacific Islands Countries typically have capacity constraints that prevent them from publishing timely high-frequency data; this project implements a simplified version of Arslanalp, Koepke and Verschuur (2021) using satellite-based vessel tracking data from the Automatic Identification System (AIS) to nowcast monthly trade flows and estimate economic disruptions due to social unrest in 2021.
- Solomon Islands is a good candidate: about 80 percent of trade is carried by sea and shipments in each port tend to be in one direction (focusing either in exports or imports).
- The project computes monthly estimators of port activity (total, by port, by activity, and inter-island), and estimates export and import volumes using AIS data.

### B. Data and Methodology
- AIS overview:
  - AIS broadcasts static and voyage-related information every 6 minutes and dynamic information every 2 to 10 seconds when moving (every 6 minutes when anchored).
  - AIS is compulsory for international commercial ships with gross tonnage (GT) of 300 or more tons, and all passenger ships regardless of size.
- Message types (as summarized in the source):
  - Static, Identity related: Ship callsign, IMO number, mmsi number, vessel name, vessel type, size, flag. Source: Manual. Frequency: Every 6 minutes.
  - Static, Voyage-related: Destination, Estimated Time of Arrival (ETA), draught. Source: Manual input by crew. Frequency: Every 6 minutes.
  - Dynamic: Vessel speed, geospatial location (latitude and longitude), heading, time. Source: Automatic. Frequency: 2-10 seconds when moving, every 6 minutes when anchored.
- Data access and coverage:
  - AIS data accessed via the UN Global Platform (UNGP) with global archive since December 1, 2018.
  - AIS data pulled monthly for January 2019 to January 2023.
- Spatial processing:
  - AIS data stored using Uber’s hexagons definition.
  - Port areas use AKV port boundaries and are covered with level 8 hexagons.
- Port-call definition and processing:
  - A port call is a vessel’s visit with clearly defined entry and exit; observations within port boundaries are pulled monthly.
  - Algorithm combines port calls that cross month boundaries and drops vessels that appear only once within port boundaries.
  - Two indicator sets compiled: port activity indicators and trade flow indicators.
- Port activity indicators:
  - Use all port calls without additional filtering.
  - Indicators computed: overall port activity, inter-island traffic, port activity by port, and port activity by economic activity.
  - Economic-activity categories: law enforcement (military and law enforcement vessels), fuel (tanker vessels), fishing, tourism (passenger, diving, pleasure craft, sailing), trade (bulk carrier, container ships, forest-product carrier, towing, tug, and general cargo vessels). Remaining ships are in residual port activity indicator.
  - Inter-island traffic counts vessels observed in more than one Solomon Islands’ port within the same period.
- Trade indicators and volume estimation (AKV-based, simplified):
  - Consider only port calls related to trade: vessel type Cargo (bulk carrier, container ships, forest-product carrier, and general cargo vessels).
  - Discard port calls shorter than 5 hours with no change in adjusted draught.
  - Estimate trade volumes based on change in utilization rate when entering and leaving port boundary.
  - Utilization rate estimation uses vessel characteristics (length L, width W, carrying capacity DWT, block coefficient) and draft information; average density of salt water used is 1.025 tons/m3.
  - Local backpropagation and historical averaging used to impute missing draft changes:
    - Local backpropagation uses incoming draft in the following port when the next port visited by a ship is within Solomon Islands.
    - Historical averaging imputes draft changes with historical average observed in the port for the ship with missing data.
- Key formulae and conventions (as presented):
  - Utilization rate at reported draft r:  
    휇휇
    푟푟
    =
    (
    퐶퐶푏푏
    푟푟
    푑푑
    푟푟
    −퐶퐶푏푏
    푑푑
    푑푑
    푑푑
    )
    퐿퐿퐿퐿휌휌
    푤푤
    +퐷퐷퐿퐿퐷퐷
    퐷퐷퐿퐿
    퐷퐷
  - Block coefficient at reported draft:
    퐶퐶푏푏
    푟푟
    = 1 − � (1 − 퐶퐶푏푏
    푑푑
    ) � 푑푑
    푑푑
    푑푑
    푟푟
    � 1/3
  - Outgoing reported draft via local backpropagation (conditional on next port within SLB).
  - Trade flow estimate:
    퐷퐷 = �
    �휇휇
    푟푟,표표표표푟푟
    −휇휇
    푟푟,푖푖푖푖
    �퐷퐷퐿퐿퐷퐷,푑푑
    푟푟,푖푖푖푖
    ≠푑푑
    푟푟,표표표표표표
    퐷퐷
    푎푎푎푎푛푛  푟푟푎푎푎푎푛푛
    ,푑푑
    푟푟,푖푖푖푖
    =푑푑
    푟푟,표표표표표표
  - Imports correspond to estimates below zero; exports correspond to estimates above zero.
- Correlation context:
  - Good imports represent 30 percent of GDP (2017-2021 average).
  - 80 percent of merchandise imports take place by sea.
  - Annual imports have a correlation coefficient of 0.94 with GDP (measured in US dollars, 1995 to 2021).

### C. Port Activity — Findings and Patterns
- Heat maps and aggregate activity:
  - Honiara port heat maps show a clear decline in activity during February 2021 and 2022 compared to February 2020.
  - Total port activity decreased by 27 percent in 2020 and by 42 percent in 2021 relative to the 2019 average.
  - Port activity in 2022 remained subdued at the 2021 depressed level but displayed an increasing trend and signs of systematic recovery.
  - Inter-island traffic broadly follows the same pattern as total port activity.
- Port-level and sectoral patterns:
  - Port activity concentrated in Honiara (main imports port) and Port Noro (main exports port), both managed by the Solomon Islands Port Authority.
  - Remaining ports display considerably less activity on average.
  - Port activity declined across all ports with the onset of the pandemic, bottoming out in early 2022 with the domestic outbreak of COVID-19.
  - A recovery trend is observed for Honiara, Port Noro, Seghe port, and Gizo Harbor; other ports remain depressed, indicating a geographically unequal recovery.
- Economic-activity breakdown:
  - Port calls related to fishing, tourism and trade plummeted with the pandemic; fuel-related port calls fell but remained relatively stable.
  - The incipient recovery in 2022 appears broad-based across fishing, tourism, and trade.
  - Recovery in fishing activity can proxy for recovery of fishing rents in the Balance of Payments; tourism recovery follows border reopening in mid-2022.
- High-frequency event analysis — November 2021 Honiara riots:
  - Heat maps for the week before, during, and after the November 2021 riots show increased water activity in the week of the riots and the week after.
  - Weekly statistics:
    - Port calls fell by 30 percent in the week of the riots (week 0) relative to the week before (week -1).
    - Trade-related port calls fell after the riots.
    - Law enforcement port calls increased during the riots period.
    - Ships that arrived the week before the riots spent on average more time in port than those arriving in the week of the riots and the week after, supporting ships facing difficulty completing tasks on land.
  - Interpretation: while port calls declined, increased in-water activity suggests ships faced difficulty unloading and increased law enforcement presence.

### D. Trade Flow Estimates — Findings and Validation
- Estimation strategies:
  - Trade flows estimated using information from all ports and using only Honiara port.
  - Validation comparisons:
    - Exports estimates validated against monthly volume of logs production and Monthly Production Index (CBSI), and compared with AKV estimates.
    - Imports validation limited; compared only with AKV volume estimates.
  - Preferred estimates:
    - Exports: preferred estimate is computed using information from all ports.
    - Imports: preferred estimate is computed using information from the Honiara port.
- Correlations (pairwise correlations for January 2019 to January 2023; AKV variables sample January 2019 to February 2021):
  - Exports all Ports vs Logs (Volume): 0.44
  - Exports all Ports vs Monthly Production Index: 0.36
  - Exports all Ports vs Exports AKV: 0.87
  - Exports Honiara vs Logs (Volume): -0.49
  - Exports Honiara vs Monthly Production Index: -0.52
  - Exports Honiara vs Exports AKV: 0.12
  - Imports all Ports vs Imports AKV: -0.03
  - Imports Honiara vs Imports AKV: 0.44
  - Exports AKV vs Logs (Volume): 0.56
  - Exports AKV vs Monthly Production Index: 0.46
- Key quantitative findings:
  - Import dynamics:
    - Import volumes show high dynamism, responding quickly to external and internal conditions.
    - Differences between the import volume estimate and the value indicator suggest increases in shipping costs and global supply chain disruptions (second half of 2021) and increases in shipping costs and global commodity prices in 2022 due to the war in Ukraine.
  - Export dynamics:
    - Exports estimated to have fallen by 64 percent in 2021 and 2022 with respect to 2019.
    - Export estimates show a large boom at end-2019 and beginning-2020 that requires validation with other data sources and authorities.
- Visual indicators:
  - Export and import volume estimates plotted against CBSI monthly merchandise value series (typically lagged by two to three months) illustrate timing and magnitude differences; six-month moving averages used for smoothing in figures.

### E. Conclusions (as per the source)
- AIS-based vessel-tracking indicators provide timely, granular proxies for port activity and trade flows that can fill high-frequency data gaps in Pacific Islands Countries.
- For Solomon Islands:
  - AIS indicators captured substantial pandemic-related declines in port activity and heterogeneous recovery across ports and activities.
  - High-frequency analysis identified disruptions around the November 2021 Honiara riots (30 percent weekly drop in port calls; increased law enforcement presence; evidence of ships unable to unload).
  - Trade-flow estimates indicate persistent export weakness (64 percent decline in 2021 and 2022 vs 2019) and volatile import dynamics influenced by shipping costs and global commodity prices.
- Preferred estimation choices informed by correlation analysis: use all ports for exports and Honiara port for imports.

*Prepared by Gabriela Cugat (RES); IMF staff estimates based on AIS data accessed via the UN Global Platform and AKV methodologies.*

### 20.      Big data can be used to fill data gaps for PICs and the IMF can serve as a

### 20. Big data can be used to fill data gaps for PICs and the IMF can serve as a capacity-building and innovation hub

### Use of AIS data for surveillance and nowcasting
- Estimators computed based on AIS data have been used as part of the surveillance dashboard by the Solomon Islands team and have been discussed with the authorities.
- The CBSI staff is exploring the use of these indicators in the context of a broader nowcasting effort under IMF technical assistance (TA).
- The current exercise showcases the potential for surveillance and TA integration in building capacity and filling data gaps.
- Initiatives like the AKV estimation exploit cross-country synergies and technical expertise available at the IMF to provide valuable inputs for both internal and external use.

### Potential single-country and global applications of AIS
- Single-country applications include:
  - Monitoring of fishing vessels to estimate fishing rents from daily vessel schemes.
  - Monitoring export-related ships to monitor for piracy/exports misreporting.
  - Tracking tourism in real time.
- Global applications include:
  - Analyzing global supply chains.
  - Assessing trade disruptions from natural disasters.
  - Evaluating the effect of trade policies.
- Given the global nature of the AIS data, it can be used to analyze both country-specific and cross-border phenomena.

### Port activity indicators and event markers used in analysis
- Port activity visualizations and indicators are derived from AIS data and IMF staff estimates.
- Figures use the following conventions:
  - Black bars represent monthly port calls of all types (commercial, tourism, oil tankers, law enforcement, etc).
  - Blue line represents the six-month moving average.
  - Horizontal lines are the yearly averages.
  - Vertical red lines represent specific events: the start of the COVID-19 pandemic (March 2020), the November 2021 Honiara riots, and the domestic outbreak of COVID-19 (February 2022).
- Heat maps and port-activity figures are presented for December for the years 2019, 2020, 2021, and 2022:
  - Honiara Port (main imports port): December 2019, December 2020, December 2021, December 2022.
  - Port Noro (main exports port): December 2019, December 2020, December 2021, December 2022.

### Implications for IMF capacity-building and technical assistance
- The IMF can serve as a capacity-building and innovation hub by:
  - Integrating AIS-derived estimators into surveillance dashboards.
  - Providing technical assistance to central banks (e.g., CBSI) to incorporate AIS indicators into nowcasting frameworks.
  - Leveraging cross-country synergies and technical expertise to produce AKV estimations and similar outputs for internal and external use.
- The exercise demonstrates how surveillance and TA integration can help fill data gaps for Pacific Island Countries (PICs).

*Source: 1slbea2023002 - 20. Big data can be used to fill data gaps for PICs and the IMF can serve as a capacity-building and innovation hub*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2023/english/1slbea2023002.pdf_