Data has always been the backbone of any actuarial analysis. From estimating mortality rates, forecasting customer demands, pricing innovative products, to making enabling distribution strategies within an effective risk framework, data-driven decision-making has remained a mainstay of the actuarial domain.
In a fast-evolving world of accelerating demands for customized products and services, rapid advancements in computational facilities, speed, and data availability, the actuarial fraternity is actively retooling itself by including data science and Machine Learning (ML) as key ammunition in their arsenal.
The appeal of Machine Learning techniques lies in the fact that they are very effective in bringing out hidden patterns in large volumes of multi-dimensional data without resorting to any major structural assumptions. The way consumer behavior might change in response to a simultaneous change in product design and shift in macro-economic drivers doesn’t have to be necessarily modeled based on a pre-determined framework; instead, a data scientist armed with Machine Learning tools and enough data can be expected to unearth the same patterns, along with more that might be buried deep in the data.
Data that tells us about a customer who has just insured her 3-story house might also give us indications on whether she needs to ensure her car, or whether she needs to top up the medical coverage for her aging parents. Machine Learning tools and technology parse such customer data far more effectively, identifying cross-sell and up-sell opportunities while optimally leveraging their firm’s distribution diversity and outreach. Many such opportunities exist in the space of risk estimation. For example, one may use consumer credit scores to correlate the customer’s financial well-being and discipline with her life expectancy.
In several parts of the world, car-mounted sensors provide real-time data on customer behavior, which can then be fed directly into the dynamic pricing of their car insurance. In the same vein, life and health insurance companies are promoting fitness bands and wellness apps tracking BMI, stress, and physical activities that can be effectively used to determine price points and risk management thereon.
Automated Machine Learning
New age techniques like video medicals and retina scans have gained momentum, which can help approve a proposal in the blink of an eye. Judicious use of such data doesn’t only improve customer risk assessment, underwriting, and product pricing, but also improves the quality of portfolio risk management as well as customer experience. Traditional modeling, while often effective with nuanced handling, can still appear hamstrung in such an unstructured multidimensional data analysis paradigm.
The advent of digital platforms has worked as a natural conduit to bring Machine Learning techniques and tools into use and harness the immense potential in data mining. Mobile apps and web interfaces provide unstructured data sources from which actuaries can glean tremendous insights. Today, almost all major websites host AI tools and bots, that generate deluges of real-time customer interaction data and transcripts which can be processed through natural language processing tools and fed into automated ML systems.
An important input in the actuarial control cycle is the detection and management of fraud. The risk of fraud has burgeoned along with the growing complexity of business, expansion of distributions channels, rising digital sales, and the ever-increasing threat of cyber-attacks. Insurers are taking recourse to Machine Learning techniques and AI tools that can identify fake IDs and impersonations, both at the issuance and claims stages.
Data mining is being used extensively to identify anomalies in policy experiences, that might be indications of fraudulent behavior. In addition, biometric data are fast becoming indispensable in KYC processes, and image-recognition tools are intrinsic to its success. The resulting seamless process is not only safer but is also creating a hassle-free customer journey and faster payment of claims within a few hours.
One must also realize that such Machine Learning algorithms become invaluable and provide incremental benefits over classical methods only in presence of a large variety of data where hidden patterns are bound to reside. Therefore, the availability of data goes hand in hand with the uptake of ML algorithms and automation. This brings us to the other critical aspect towards the adoption of ML – the easy availability and financial viability of data storage and computing platforms.
Many Machine Learning techniques and supporting software programs which have been around for a while are becoming popular only recently. Major strides have happened in cloud computing, and market movers like Microsoft and Amazon are striving to make their cloud platforms cheaper and easily accessible. This means that even smaller players like InsureTech and FinTech selling insurance and financial services through apps can now tap into such capabilities without having to make a huge investment in building a supporting IT infrastructure
The actuarial profession and premier education institutions focusing on applied statistics and specialized actuarial courses have started to adapt to this evolving demand. Besides integrating data science and Machine Learning in the syllabus for new actuaries, the Institute of Actuaries in India and abroad have been conducting webinars and hands-on practical sessions to encourage uptake of Machine Learning and upskill the existing workforce.
College curriculums now include data science modules and programming languages where students are trained to handle big volumes of data and be conversant with state-of-the-art software available in the market. The net result for an insurance company is that it is easier for them to put together a team of data scientists that can meet their requirements; extract insights and enhance product suite and services.
Machine Learning has caused a stir in the actuarial fraternity and opened new avenues of data-driven innovations and risk management. It has increased the relevance and reach of actuaries beyond mainstream insurers and reinsurers. Insurtech and fintech firms are employing actuaries to help them build newer products and financial solutions and create the framework to service and sustain them. Clearly, this ML journey is destined to show the way for actuaries to evolve, grow and keep contributing.
Chief and Appointed Actuary