The modern history of global trade will be divided into B.C. and A.C. — before-corona and after-corona. What happened B.C. might as well have occurred in an alternate universe and the pace of change A.C. is so frenetic that even one-month-old data has lost much of its value.
With make-or-break business decisions in the balance, the focus has predictably turned to data that signals future trends, such as container-shipping "blank sailing" information from companies such as eeSea and Sea-Intelligence, and real-time "nowcasting" data on ship movements from companies such as CargoMetrics.
There is a new entrant in the ocean-shipping nowcasting space: the International Monetary Fund (IMF). An IMF team made up of economists Diego Cerdeiro and Andras Komaromi and data scientists Yang Liu and Mamoon Saeed published their initial proof-of-concept findings on Thursday.
"With the current fast-changing developments, policymakers need to know what is happening to the economy in real time, but they often must settle for data telling them what happened many weeks ago," wrote the team.
The solution, they believe, is "daily monitoring of trade developments in real time [to] help provide a reliable early warning regarding potential economic contagion effects amid the pandemic."
The IMF system estimates global trade volumes expressed as indexes of 30-day moving averages for 2020 versus comparable 2017-19 averages. The data is being used to develop the IMF Global Trade Intelligence Index.
Chart credit: IMF. Source: Cerdeiro, Komaromi, Liu and Saeed (2020). AIS data from MarineTraffic. Note: 30-day moving averages, based years adjusted for Lunar New Year
The first chart published Thursday covered Chinese export volume and clearly shows the extreme collapse in February as a result of the initial coronavirus shutdown in Wuhan.
That supply-side shock was followed by a resurgence in activity through March and early April as China's export network reopened. The IMF data then shows a downturn in export volumes as a consequence of a demand-side shock: European and U.S. imports falling in the wake of social-distancing measures.
According to the IMF estimates, Chinese exports started the year around 10% higher than average, fell to 30% below average, rebounded to near average, then fell back to 10% below average.
Global segment volumes
The IMF team also published global import and export data broken down by cargo segment: dry bulk, oil and chemicals, containerized goods, and cars.
The first important takeaway from the segmented data is that bulk goods, whether liquid or dry, appear to be performing better than unitized goods, i.e., containerized goods and vehicles. Bulk goods are largely sold to industrial entities, unitized goods to consumer-focused businesses.
Chart credit: IMF. Source: Cerdeiro, Komaromi, Liu, and Saeed (2020). AIS data from MarineTraffic. Note: 30-day moving averages; the methodology is only designed to track trade in goods
Both tanker and dry bulk volumes remained average to above-average in relation to previous years, at least in the period covered by the IMF data. In the case of dry bulk, rates for larger bulkers collapsed this week, beyond the time frame of the IMF-published data (although that data does show dry bulk imports turning negative in late April and early May).
On the tanker front, it has been widely reported that more laden ships are at sea, either because of floating–storage contracts or because of the inability to unload due to lack of land-based storage.
The IMF data is consistent with this dynamic. The export and import trend lines diverge, with the export line continuing to rise since February and the import line on a slow decline since mid-March.
The IMF team explained, "In the specific case of oil and related products, the recent export performance is especially strong but is not fully matched by an increase in world imports, in line with reports that crude oil is being stored."
Another key takeaway from the IMF data is that the global vehicle trade is performing dramatically worse than any other sector. Global vehicle export volumes plunged by more than a third during the month of April versus the 2017-19 average, which the IMF team attributed to "companies across the supply chain halting production."
The IMF data also provides insight on container volumes, an important indicator for U.S. trucking and rail sectors, as well as ocean carriers.
Global box exports fell below average in February, rose above average in March, then fell to 10% below average by late April. Container imports were at average to slightly below-average levels in March and rose to almost 5% above average levels in April.
As previously reported by FreightWaves, import volumes rose in April predominantly due to cargo ordered immediately following the end of the Chinese lockdown but before the U.S. lockdown suppressed U.S. cargo demand.
Because of the time lag between container exports and imports as ships spend two to four weeks in transit between Asia and the U.S., plus additional days for containers to make their way through the terminals for pickup by trucks and trains, the effects of declining exports from China last month are starting to be felt in the U.S. this month.
The IMF team authored a working paper released on Thursday, "World Seaborne Trade in Real Time: A Proof of Concept for Building AIS-based Nowcasts from Scratch," which details the methodology behind their system.
The team used the Automatic Identification System (AIS) ship-positioning data from MarineTraffic sampled hourly since Jan. 1, 2015. Over a billion AIS signals have been inputted into the IMF system.
The team used the AIS data to build a port dataset, machine learning to imply export versus import ports, and data on ship capacity and assumptions on whether ships were laden or in ballast to estimate volumes. The team then benchmarked its system's output against official trade volume data; the benchmarking correlation was 0.88 for monthly global volumes.
"Our proof of concept clearly demonstrates it is possible to build a real-time global trade nowcasting system relying only on raw AIS data and off-the-shelf machine-learning algorithms," said the team.
It is a work in progress. The next steps: Refine the algorithms; improve forecasting properties by predicting ship destinations while ships are en route, as opposed to the current method, which counts them when they arrive; incorporate information beyond AIS data to improve cargo weight estimates, and bring shipping experts into the fold to prevent double-counting and other errors.
The most important question for cargo shippers is: Will updated data from the IMF Global Trade Intelligence Index be consistently available to the general public, or will it only be used internally by the policymakers of the IMF?
In response to a query from FreightWaves, Cerdeiro responded, "We are investigating what could be shared externally — what aggregation level and frequency — but no decision has been made for now." Click for more FreightWaves/American Shipper articles by Greg Miller
Photo credit: Flickr/Tom Diggers