Machine Learning, Artificial Intelligence and CO2 Reductions

Shipping folks on my Linked-In and Twitter feeds were all abuzz in the past week with a posting by a Professor at the Norwegian School of Economics (NHH), who had spent some time in Boston, regarding drybulk chartering. Specifically, the post hinted at yet-to-be published results of a study where machine learning was used to derive decisions on fixing of Capesize bulk carriers. While renewed activity on the iron ore front has led to a recent increase in Capesize hires, the results hinted at from the NHH suggest a bump up of 10%, or around $1,700/day, versus just “taking the market” during 2017 – 2019- when the Baltic Capesize Index was worth around $17,000/day, on average. OK, in a 2007 or 2008 style hypermarket, $1.7K/day is a rounding error- if that. But in 2020 style markets- such gains or worth a look. I am reminded also that a product tanker operator (tied to a large Danish company, and recently featured in a Wall Street Journal article) in conjunction with some different Boston based quants, alluded to a similar advantage at a big maritime conference (when they were still live, not virtual), using an artificial intelligence approach for choosing whether to send their vessels eastbound, or westbound (looking at oil supply patterns in addition to ship movements). The conference speaker indicated that advantages, the heftier TCE’s, were conferred for months- not for years outward.

The comments on Linked-In regarding the NHH project were very telling, revealing the divergence of views on one hand- about the efficacy of such efforts (or whether they were worthwhile- at all) to those who wanted to see more information about it. There were some hints that the algorithms in the background, driving routing choices on taking Brazil / China runs versus Australia to China trips (not a whole lot different from the east versus westbound tanker routings), use macro data but also highly specific commodity data. A group of commentators on Linked-In waxed positive but said that they would  reserve judgement on the release of a detailed write-up from NHH.

It’s interesting that in both cases, the analysis did not focus on predicting the rises, or falls, in the overall drybulk or tanker spheres, respectively. But in both cases cited here, where assets can enter or exit particular trades at will, a micro-economic demand side view (on particular trades) opens up fresh insights- and maybe (secretiveness is part of the fun here) economic advantages.

So, forget about Stamford, Connecticut or similar spots; Boston really seems to be where it’s at. Yet another Boston based group, now officially a “stalwart” since its inception in the mid 1980s when models were still crunched in Lotus 123, has been looking at the dollar values of investments in “ESG”, in conjunction with a prestigious university up there with a rich shipping tradition. This is a huge challenge, bringing macro concepts, like cutting carbon emissions is worth $<pick a number> trillions, down to the analysis of individual vessels. Yet, the lesson of the super duper economic models could be that disaggregation may be the way to go.

There’s been a lot of talking about reducing CO2 emissions, and even getting them down to naught. There are discussions about fuel tax levies (did I hear $200/ton from an environmentalist at a recent maritime conference?), carbon trading schemes and the like. Thinking about optimizations- where there’s been a lot of talk, and Artificial Intelligence (if it really works), maybe some Bostonians (or Norwegians) could think about tying reductions to increased earnings on particular trades – with blue chip ESG-minded charterers “paying the freight”. This seems like an opportunity that should not be wasted.