Małgorzata Śmietanka and Philip Treleaven explain the technologies making up the data-driven insurance picture
The data science technologies of big data, the Internet of Things (IoT), artificial intelligence (AI), behavioural or predictive analytics, and blockchain are poised to revolutionise the insurance marketplace and create a new generation of insurtech companies. In China, Zhong An (a digital insurance collaboration between Alibaba, Tencent and Ping An) underwrote more than 630m insurance policies and serviced 150m clients in its first year. We are undergoing a data revolution led by multinational companies such as Amazon, Google, Facebook, Uber, Alibaba and Tencent. This is radically changing the way we work, socialise and buy products or services. The impact of Amazon on retail, Uber on transportation and Netflix on entertainment shows that whole sectors can be transformed or even disappear – but this also brings opportunities. The future of legal services is an interesting example – for example computable legal contracts will enhance efficiency and quality while reducing transaction costs. Maybe an ‘Amazon-for-insurance’ will emerge to dominate insurance, and it might be Chinese.
Tencent’s WeChat, which is similar to Facebook’s WhatsApp, provides a glimpse of the future. WeChat is on course to disintermediate retail and financial services in China, and is expanding into Africa and South America. China is considered to be years ahead of the rest of the world when it comes to social messaging. Hail a taxi, split the bill, buy a cinema ticket, invest in financial products, read the news, follow brands, transfer money to friends, shop online, meet new friends nearby – all this and more is already available to Chinese people within one application. Tencent thus possesses vast amounts of user data, allowing precision profiling of individual customers for the purposes of, for example, future insurance risk profiling – should users consent. This presents it with the opportunity to enter new markets.
What are the key data science technologies, and how do they interrelate?
Big data consists of two things: harvesting huge amounts of business and social data, and examining these very large datasets to uncover hidden patterns, customer preferences and risks. The insurance industry already collects enormous volumes of data and thus has major opportunities for analytics. This allows precise risk profiling of individuals and the building of complex ecology models for long-term predictions of, for example, natural disaster impacts and national health crises.
Internet of Things
IoT is the inter-networking of ‘smart’ physical devices, vehicles and buildings. Networking enables these objects to collect and exchange data, and contribute big data. As more devices come with an internet connection, insurance companies can manage risks in real-time, assuming consent and privacy issues are addressed.
AI technologies, especially machine learning, are increasingly used to analyse data, interact with customers and act as intelligent assistants for professionals. Consumer examples include Apple Siri, Amazon Alexa, ‘robo’ advisers and autonomous vehicles.
AI allows computers to learn dynamically and make decisions. There are two main branches. Knowledge-based systems are computer programs that make decisions based on knowledge represented by a decision tree. Machine-learning algorithms can perform a specific task without explicit programming, relying on patterns that arise when the machine is exposed to new data.
Two increasingly important AI applications are natural language processing and behaviour analysis. The former is the application of computer techniques to the analysis and synthesis of natural language and speech, while the latter involves computationally identifying and categorising opinions expressed in a piece of text.
Predictive analytics involves extracting information from historical and real-time datasets in order to determine patterns and predict future outcomes. It can forecast what might happen in future, and can help with ‘what if?’ scenarios and risk assessment.
Blockchain technology, originally conceived for the cryptocurrency bitcoin, has far-reaching potential in other areas, especially business. The core technologies are distributed ledger technology (DLT) and smart contracts. DLT is a decentralised database where transactions are kept in a shared, replicated, synchronised, distributed bookkeeping record, secured by cryptographic sealing. The distinction between distributed ledgers and distributed databases is that the nodes of a distributed ledger are assumed not to trust other nodes, and must independently verify transactions before updating the ledger. Smart contracts are simply programs that codify transactions and contracts, with the intent that the records managed by the distributed ledger are authoritative with respect to the existence, status and evolution of the underlying legal agreements they represent. Smart contract technology has the potential to automate laws and statutes.
Key applications for insurance
Three key data technologies currently driving change in the insurance industry are: AI-based analytics allowing improved underwriting; computable insurance contracts processed automatically; and blockchain-based digital marketplaces.
AI-based analytics will help insurance companies to gain new business by allowing them to personalise insurance products and better respond to clients’ needs. For example, some insurance contracts for those with type 2 diabetes can be three times more expensive, no matter how well that condition is managed by the insured. However, more accurate underwriting models and constant learning from new data could be reflected in premiums.
Computer-readable and executable legal specifications are set to have a profound impact on business and legal services. So-called computable contracts are legal specifications that a computer can read, understand, verify and execute – and therefore automate. The challenge is to specify a contract that can be composed and read by professionals, and can also be translated into a domain-specific computer-readable specification such as XML. Here, the big potential is automating high-volume, low-cost domestic insurance without human intervention.
As insurance services become globalised, there is a major opportunity to create an Amazon/Alibaba model for digital insurance services using blockchain technology. Blockchain marketplaces are under development in Singapore and Dubai. However, the most comprehensive digital blockchain infrastructure programme is Estonia’s e-Estonia, where every citizen has a digital identity, digital signature and personal record, and virtually all government services are digital and online.
Other insurance opportunities
There are also advantages for claims reserving and behavioural change programmes.
Individual claims reserving is gaining popularity across many actuarial associations and will most likely replace current models based on aggregate data. The claims reserve is the largest number on the liability side of the balance sheet of a typical non-life insurance company, and reserves are essential for the financial strength of the company. Machine learning may help to improve the accuracy of claims reserving, and individual claims features can lead to model improvements and more accurate risk assessment.
Behavioural change programmes are increasingly popular, as insurers become more involved in policyholders’ health. AI analytics help in monitoring health markers, making use of data collected by wearables such as smart watches and sport apps. This information could be reflected in the premium paid by a policyholder.
Embracing the opportunities of data-driven approaches will help insurers improve risk management, marketing, reserving and claims handling, among other areas.
Małgorzata Śmietanka is PhD researcher in the UCL Computer Science Department, leading the Actuari project at UCL and insurance industry collaboration through UCL spinout Actuari Ltd
Professor Philip Treleaven is director of the UK Centre for Financial Computing and Business Analytics, and professor of computing at UCL.