Marine insurance has always depended on data. Accurately predicting, evaluating, and pricing risk fundamentally relies on access to good quality data. This was just as true in the 17th century as it is today. When Edward Lloyd opened his now eponymous coffee house, he understood the value of data, providing intelligence briefings to the merchants, shipowners, and insurers who frequented his establishment. By 2020, it is estimated that 1.7mb of data will be created every second for every human on earth. With a wealth of data now at its fingertips, how is the marine insurance industry making use of it?
What is big data?
Exactly as the name suggests, big data is the process of computationally analysing extremely large datasets. Usually, this involves processing both structured and unstructured data and gaining some kind of valuable insight from it. The world’s commercial fleet now produces billions of data points each year. OEM machinery data, AIS broadcasts, cargo tracking sensors, trade documentation, and even CCTV systems create information that can be used to create business value. If you want to read more on exactly what big data is and how it is being used in maritime, click here.
How is big data being used in insurance?
In insurance, the business value that can be created through data is huge. The more data an insurer can process, analyse, and derive insight from, the better risk management and claims handling processes become.
Broadly speaking marine insurance is split into two camps Protection and Indemnity (P&I) and Hull and Machinery. P&I is a form of mutual assurance, where shipowners collectively pay into a fund which covers losses not insured through traditional insurance. P&I is provided by not for profit “Clubs”, whose membership is made up of the shipowners they insure. P&I generally covers third party liabilities that most traditional insurers are reluctant to cover. Hull and Machinery, on the other hand, works more like traditional insurance and covers the risk to a ship owner’s primary asset, the ship and its machinery.
Where data is concerned the type of insurance is irrelevant, it is still applied in the same way; to price risk, apportion liability, and calculate claims to injured parties. While there is no doubt that the proper application of data science will transform the entire insurance industry, there are three prominent application areas emerging in maritime: smart premiums, claims handling, and loss prevention.
There are a huge number of factors that need to be taken into account when pricing risk and calculating the premiums for a ship going to sea. Is the ship seaworthy? Are the crew competent? Where will the ship be sailing? What will the weather be like? What cargo is being carried? Broadly speaking, insurance premiums are a blunt instrument, designed to cover the widest possible risk for a set period of time. Insurance needs to cover a broad range of potential situations, from sailing near a hurricane to transiting a narrow channel.
Usually, a ship will be able to sail within the bounds of its insurance policy, with restrictions in place for certain geographies or cargo types. This approach can be restrictive, requiring vessel operators to either pay for expensive policies that cover an incredibly broad range of scenarios, manually request policy adjustments based on charterer demand, or to only engage a vessel in a specific type of trade.
With enough data, it is possible to accurately profile a vessel’s risk level on an ongoing basis. By combining historical vessel, cargo, and claims data with current tracking and weather data it is possible to dynamically and automatically calculate and price risk. This approach makes it possible for insurers to introduce smart products where premiums are adjusted dynamically to minimise cost and maximise coverage. This gives ship operators the ability to charter their ships with more flexibility, allowing them to pursue more commercial opportunities. It also minimises administration expenses for both insurers and ship operators, with benefits that compound when entire fleets are covered under a single, dynamic policy.
Accurately processing insurance claims requires a forensic approach to incident investigation. In any marine loss, there can be multiple parties involved; usually, there are multiple parties facing some sort of liability and many more who have a claim. It is often difficult to establish the causes of a loss, with marine claims often having to be settled by the Admiralty Courts in a process that can take months if not years. Where possible, using data to reconstruct an incident to quickly establish exactly what went wrong can speed up the claims handling process, making the settlement process faster and less expensive.
Cold chain cargo is a great example of a space where better data visibility leads to better claims management. By collecting temperature, position, and humidity data when cold chain cargo is in transit it is possible to automatically detect whether the cargo is safe to use on delivery. If the cargo is spoiled, it is possible to know exactly when the temperature exceeded safe limits, who was responsible, and which parties are liable for the damage.
While refrigerated logistics may be clear cut, vessel collisions and allisions are not, with liability usually apportioned between all of the vessels involved. With the right data collection and processing tools in place, it is possible to create a digital twin of a collision incident using data from AIS, ECDIS, radar, bridge VDR, and video from available sources. Once a simulation is created it can be run countless times to help investigators test different hypotheses of what went wrong and what could have been done to prevent the incident.
As well as helping to apportion blame, simulation tools are incredibly powerful when it comes to learning from incidents to prevent future losses. Due to the nature of the risks they insure, and the fact that they are mutual, not for profit organisations, P&I Clubs are particularly interested in preventing future losses from past incidents.
At this point, we begin to cross the divide from insurance to ship management and operations but there are some promising applications coming to the fore that bridge the gap. Claims data from P&I has always been used by loss prevention teams to establish new guidelines and best practices, but now it is possible to use that data to feed risk management systems that flag up dangerous situations before they occur. If properly implemented these systems can minimise losses and maximise safety on board for a fraction of the cost of traditional loss prevention methods.
Startups and funding in the space
This is a relatively mature sector for the maritime startup industry, it has been clearly influenced by the more well known and well funded InsureTech space. Nearly half of companies operating in the space have received venture funding of Series A or above.
$127million has been invested in data startups aimed at the marine insurance market since 2014. Only small amounts of venture investment were made between 2014 and 2016, but $92.5million has been invested since 2017 with Israeli startup Windward leading the pack following a $16.5million Series C round in 2018.
Notable startups and use cases
Winward uses data to help people understand maritime risk and take action to reduce it. Their Vessel Operational Profile uses behavioural and ship characteristic data to benchmark vessels against hundreds of dynamic risk factors. This allows insurers to better asses their own portfolio risk, making it easier to rapidly and safely underwrite assets at sea.
Parsyl uses small inexpensive IoT devices that can be shipped along with high value or sensitive goods. The system gathers data throughout the journey and automatically turns sensor data into actionable reports and visualisations. The data can be used to quickly and easily handle claims and, over time, to develop smart risk profiles of particular shippers, carriers, or cagoes.
Geollect specialises in using geospatial data to help clients build up operational intelligence and understand risk. They are work with leading P&I clubs to build tools to collect and display incident and near-miss reports from vessels around the world.
The Concirrus Quest Marine product accesses and interprets large sets of static demographic and dynamic behaviour-based data set. Combined with historical claims information they can reveal the behaviours that correlate to insurance claims. The outcome is new insights and rating factors that simply did not exist before, the ability to better deploy risk capital, improve loss ratios and drive down operating costs.
While big data has an important role to play in analysing insurance risk, it is part of a much wider technological picture. These vast data sets, when combined with artificial intelligence, process automation and blockchain could unlock untold value when offered as part of a smart insurance service that is dynamically priced and offers automatic payouts based on data defined rules.
Though more mature and better funded than many other ShipTech sectors, there are still huge opportunities in the marine insurance market. As more IoT services come online, and live operational datasets become more widely available, the opportunity to connect to ships directly to build new insurance products will only become greater.