Data-centric Insurance: How the London Market can embrace analytics and regain its data crown

Transcription

1 Data-centric Insurance: How the London Market can embrace analytics and regain its data crown

2 2

3 Introduction The London Market has a long tradition of excellence in the actuarial analysis of premiums and claims data, but it has not yet embraced the analytics revolution seen in other industries. For the London Market to remain globally competitive, it must re-establish its pre-eminence regarding the use of data. The London Market has historically been a pioneer in the collection, manipulation and utilization of data. Over time, insurers have built up vast quantities of proprietary data and developed proven methodologies to price risk and process claims. However, the pervasiveness of the digital revolution has meant a wide number of other industries now see data-driven business decisionmaking as core. These industries have overtaken insurance to become leaders in advanced data science and insight generation. They harness the potential from the rise of the Internet of Things (IoT), the availability of third-party datasets and the proliferation of unstructured data sources. To do so, they leverage scalable, powerful data-analysis technologies including machine learning and AI to create actionable business insight and enhanced customer value. While other industries reinvent themselves, the London Market itself faces significant additional challenges. The continuing macro-economic conditions combine a soft market with low interest rates, and dampen profitability of the industry as a whole. The cost of transacting insurance business in London remains high, and the market modernisation agenda while progressing will only deliver benefits in the medium term. The London Market s share of business from emerging markets is in decline as that business is increasingly written in local markets. Turning from these challenges to data itself, the London Market has its own unique issues to address. The insurance business generates vast quantities of data, ranging from highly conformed messages to content-rich but unstructured reports. This raises two difficult questions - how to ingest this data in a meaningful, costeffective and timely way, and how to maximise the utility of data in the Market while allowing owners of the data to retain their competitive edge. Analytics can be a critical component of the strategy to respond to these challenges, and can help both carriers and brokers develop new opportunities for growth and improved profitability. Analytics has the potential to dramatically improve core competencies such as risk pricing and claims management. However, it can also enhance risk monitoring and risk prevention, improve end-customer segmentation and understanding, and help drive operational efficiencies. It is time for the London Market to embrace the analytics revolution and regain its data crown. However, in order to do this, the Market and its participants will need to adopt the right talent, culture and technologies to overcome existing barriers. 3

4 Why invest in analytics now? Those carriers and brokers who place analytics at the heart of their digital strategies will be able to reinvigorate product innovation, increase their profitability and meet developing regulatory requirements. Drive innovation The Commodity Cone illustrates the importance of product innovation for the London Market. Innovation occurs as emerging risk classes are identified and corresponding products created, and this depends on underwriters being able to identify and price risks, often with very limited information. Carriers and brokers can use advanced analytical techniques (such as machine learning) in the pricing of emerging risks, based on new types of internal and external data. By doing so, analytics can help drive and accelerate product innovation. More innovative use of analytics-driven pricing may emerge. Crowdsourced risk models may become more effective and adaptable than proprietary models, thereby lowering the barriers to entry for writing Specialty business. Usage-based pricing models where premiums flex daily in line with a changing, monitored risk profile may eventually be used for commercial property or marine business. The rise of the IoT, coupled with advanced analytics, allows carriers and brokers to focus more on prevention of loss than on simple compensation after the event. The huge wealth of data now available makes real-time preventative advice a possibility. London Market Commodity Cone New Risk Coverages and Service Innovations Emerging Mature Commodity Innovation and Reinvention of Existing Coverages and Services Improve profitability in a soft market Looking at the more mature risk classes, our experience indicates that the proper application of analytics could increase carrier profitability by up to 11.5%. Manage Risk & Pricing Pricing and portfolio management can be improved through the combination of advanced analytics fed by larger amounts of data. Better Risk Monitoring Enhanced Portfolio Management Improved Underwriting & Pricing Reduce Operational Costs Operational costs can be reduced through intelligent automation opportunities that are introduced as initiatives such as PPL continue to support the Market s move away from reliance on paper. Improved Claims Assignment Improved Policy Admin Efficiency Optimised Service Cost 1% 2.5% Up to 11.5% profit growth 4% 4% Reduce Claims Loss & Prevent Fraud Fraud analytics (2.5%) and intelligent data-driven decision support for claims processing (1.5%) can together generate savings of up to 4% on claims costs. Improved Claims Processing Analytics Improved Subrogation/Recovery Improved Compliance Fraud Analytics Fraud Identification & Mitigation Improved Internal Audit Mechanisms Increase Client Retention A data-driven approach to client retention and pipeline management can identify must keep profitable risks while flagging the business at risk of being lost. Reduced Attrition Value Added Risk Management Improved Client Experience & Service 4

5 Respond to regulatory challenges Financial regulators are placing greater emphasis on the use of data and analytics. These regulatory pressures will continue to develop, and require insurers and brokers to adopt an effective data management strategy. Regulators are increasingly keen to understand the implications of analytics for customer protection. Solvency II requires insurers to have a better understanding of both the source and accuracy of the data flowing through their organisations. More recently, the FCA launched a process to understand how retail general insurers use big data and how this impacts consumer outcomes and competition. Regarding London Markets in particular, regulators have also highlighted specific data-related challenges. The FCA thematic review of outsourcing found significant deficiencies in the management information (MI) it considered necessary to discharge the obligations of the insurer and intermediary regarding the management of Delegated Authorities. Respective regulations for Sanctions Management and KYC / KYA have emphasised the need for insurers to better understand the nature of the original insured. 5

6 What does a more modern analytics business look like? Advanced analytics differs from traditional actuarial analysis because it uses both internal and external datasets, it is able to identify new predictive models, and it can provide real-time actionable insights to business users. Data for analytics Traditional actuarial analysis is built around batch-type processing of internallyheld transaction and risk data. It uses predefined models to process and analyse structured, managed datasets of historical data in order to determine rates and prices. A modern approach to analytics would look radically different. Multiple external datafeeds (such as those relating to commodity prices, social media or weather) could supplement the core internal datasets for the various classes of business. Unstructured data held internally (e.g. claims notes or risk reports) could be analysed semantically for recurring keywords that might indicate a new trend or a red flag. These additional datasets can often be high-volume, highly variable in quality and of time-bound value. As such, the technology solution requires burst capacity for processing and storage, of the type that may be provided by cloud infrastructure. Multiple data sources can be consolidated and managed in data reservoirs using technologies such as Hadoop. Analytics models are prescriptive, not predictive Actuarial teams build statistical models that analyse structured datasets to find new associations and correlations that can be used to predict outcomes. Analytics models, by contrast, can produce prescriptive results identifying the Next Best Actions for underwriters and claims handlers that may suggest perils for exclusion on a policy, or support appropriate claims assignment. For more commoditised risk classes, cognitive underwriting with machine learning developed to a point where underwriting decisions can be made from disparate datasets can more effectively price risk if the relevant datasets are too diverse to be assimilated by one person alone. Analytics is for the end-users Advanced analytics is focused on giving actionable business insight for decision making direct to the business user. Techniques such as visualisation are used to deliver meaningful outputs that can be directly interrogated by the business user. However, the mere adoption of new tools and techniques will not in itself deliver the full benefits of advanced analytics. The modern analytics culture embeds data-driven decision making throughout the whole insurance and broking value chain not just the traditional data areas of Actuarial. It is based on a philosophy of continuous testing and learning, making adjustments to the approach based on client feedback, and prepared to fail fast where ideas and approaches do not deliver valuable, actionable business insight. 6

7 Why have London Market participants been slow to adapt? Five long-standing orthodoxies have dissuaded participants from adopting advanced analytics. Data ownership prevents the establishment of centralised analytics As data flows through the London Market, it is handled by many parts of the value chain. Even though the data can be seen to originate with the client, there are differing and strongly held views of which data is owned by whom. This conflict can discourage organisations from making full use of the data they receive as part of doing business within London. Moreover, because of the competitor/ collaborator nature of the London Market there can be a fierce rejection of any sort of shared market wide data sharing or analytics. This thorny question of data ownership will have to be resolved to allow organisations to take full advantage of the data and drive new insights using analytics. Companies are swamped with increasing amounts of data & will not be able to analyse it at scale Each participant in the market receives large volumes of risk, exposure and claims data at an increasing rate. This data can be structured, conformed data or unstructured data. There is a perception that the quantity and variety of data is so huge that it is not possible to use even the conformed data to derive value from analytics. However, the maturity of super-high capacity tools such as Hadoop makes it possible for companies to manage all of their data structured or not and apply analytics to it at pace. Legacy systems are too expensive to upgrade In a market overburdened with legacy systems, the benefits of technology refresh have often been outweighed by the cost of redeveloping the custom components. The compromise has often been to wrap new functions around core legacy systems, resulting in variable data quality. Analytics-as-a-service can change this. Data analysis belongs in the back office The traditional statistical approach to evaluating loss reserves and developing rating models is seen as a back-office function that does not interact daily with the underwriting business. Advanced analytics, on the other hand, could prove invaluable to the front office. Underwriting is an irreplaceable skill Those who see data-centricity as a threat to the more personal interaction- and intuition-based approach to underwriting are stalling the adoption of analytics in supporting front-office decisions. 7

8 Where can advanced analytics deliver value? Once investment has been made in advanced analytics, it can be used to enhance value across all aspects of the business and value chain, from initial placement through to risk management, compliance and claims handling. Furthermore, it can also be used to drive product innovation. Case Study: The Connected Shipping Fleet Driven by the Internet of Things, the connected ship uses a network of sensors built into new vessels to give ship owners near real-time access to an increasing range of information concerning their fleets. Types of data feed available include location, weather and ocean current data, as well as the state of on-board equipment and cargo data. The implications for Marine insurance are significant. Real-time knowledge is now available about factors such as proximity to pirate activity, or ports experiencing loading/transit delays. This will allow insurers to monitor their risk profile more accurately and allow the insured to better manage scheduling with third parties. Container sensors can now transmit enhanced data, such as stacked height, temperature, moisture content or even location if lost overboard. Armed with this data, insurers may be able to recover lost containers, or indeed prevent other ships hitting them. Ultimately, the Connected Ship offers both carriers and brokers the opportunity to better understand and mitigate the risk the insured seeks to indemnify. Case Study: Assessing the likely profitability of large commercial PD/BI programmes Assessment of overall risk for attritional (non-natural catastrophe) exposure for Property Damage / Business Interruption business is difficult for large commercial clients, due to the complexity of coverage and the range of exposures contained in a single account. Up to now, the approach has been to commission engineering reports on a sample of locations and scale up the results to predict the risk profile of the entire population. This method has been shown to be ineffective in distinguishing between accounts that are likely to be profitable or not. By contrast, a business relevant model, designed in collaboration with MIT, applied advanced analytics to client data and predicted profitable and unprofitable risks with an accuracy of more than ninety percent. This approach helps underwriters to select profitable business more accurately and more quickly. 8

9 Further potential use cases Enhance risk management and provide near-real time risk advice by analysing data from IoT devices embedded within property and high-value assets Augment client risk data with publicly-available information to meet the fully cognisant requirement of Insurance Act 2015 Objectively assess the performance of loss adjusters prior to instruction, to allow fact-based selection Analyse driver data from multiple vehicles in order to identify high risk behaviours across the fleet Use intelligent analysis of claims data to proactively alert claims handlers when there is an increased likelihood of claims leakage Automate basic compliance and coverage checks using semantic analytics, allowing wordings experts to focus on adding value in specialist cases where needed Analyse machinery maintenance reports in order to evaluate the attitude of policyholders towards maintenance and predict the probability of machinery breakdown Create a market-wide shared service to provide competitive advantage for London Market participants through a continuous flow of consistent data and access to advanced analytics-as-a-service Identify types of employees who pose a greater cyber risk to the client by analysing publicly-available social media data, and develop improved education programmes to mitigate risk 9

10 How can the Market move forwards? Some London Market participants are taking steps to adopt an analytics culture. Getting started can be simple. Understand the existing internal data estate Carriers and brokers can start by mapping out their internal data flow and asking themselves some key questions to test understanding of their existing data estate: How well do we understand the end-to-end data flow, from client to broker to (re)insurer? Are we sufficiently leveraging all of the internal data we hold? Is our data centralized or held in silos? Are we using our unstructured data effectively? Where are the key gaps in our data, where is this missing data needed, and how can we address this? Identify the next generation of Data Scientists Data Scientists are central to the successful adoption of analytics, and businesses should act now to identify potential candidates. The ideal London Market Data Scientist has the following characteristics: Underwriting and Claims experience with a London Market perspective A rigorous approach to, and passion for, data An engineering mindset, including knowledge of statistics and algorithms Practitioner of Hadoop and other BI technologies Other viable candidates may only have some of these characteristics, but the key to success will be their capability to acquire the remainder. Establish the target technology platform Leveraging the right tools and technologies is critical to maximising the potential of big data for London Market participants. It is vital to design a suitable data strategy and platform architecture prior to adopting analytics at scale, such as the following illustrated key components. Visualization best-of-breed self-service BI and geospatial visualization Analytics building blocks for modelling & processing analytics applications Storage cloud architecture and other validated storage Data Ingestion integrates data preparation and extraction tool Additionally, capital investment can be minimised through the use of analytics-asa-service, rather than having to establish these components in-house. Define and launch the new culture Analytics should become a core business competency, and its success should be measured on its ability to deliver actionable business insight. Its leaders and advocates should come from the core parts of the business and simply be enabled by technology. When it comes to launching analytics projects, these should initially be managed in an environment that is as greenfield as possible rather than simply being an adaptation of existing MI / BI solutions. Furthermore, to avoid being constrained by the legacy of the organisation from a technology, cultural, and governance perspective the delivery team should be seconded from the business and freed from their day-to-day BAU responsibilities for the duration of the project. Begin the experiment and target repeatable quick wins The happy path for analytics deployment is building prototypes as proofs of concept, followed by a limited pilot and eventual industrialised deployment. The reason for the prototyping is to test specific hypotheses about how to generate better business insight. Senor business stakeholders will lead the generation of ideas which will form a pipeline of hypotheses to be tested in a rolling programme of individual proofs of concept, each lasting eight to ten weeks. This is sufficient time to understand whether there is demonstrable business value in the hypothesis. Hypotheses that are not validated are allowed to fail fast. Successful hypotheses will be developed into a pilot to deliver actionable insights for selected areas of the business. 10

11 Conclusion London Market participants, and indeed the Market as a whole, can gain a significant competitive advantage by applying up-to-date advanced analytics to currently available data. Despite broad recognition of this fact, many participants are held back by legacy systems and a belief that traditional actuarial approaches can deliver the same outcomes as advanced analytics. In reality, these advanced analytics capabilities can be applied at pace on scalable platforms, or analytics-as-a-service, to deliver a wide range of benefits throughout the business. Gearing up to use these capabilities, carriers and brokers can deliver near-real time, actionable insights directly to their decision makers throughout the business, thereby unlocking the true potential of their data. 11

12 Contact the authors Max Richter Managing Director, UK Insurance Analytics Lead Jamie Althorp Managing Director, London Market & Specialty Insurance Lead Peter Temperley Senior Manager, UK Insurance James Thomas Senior Manager, UK Insurance About Accenture Accenture is a leading global professional services company, providing a broad range of services and solutions in strategy, consulting, digital, technology and operations. Combining unmatched experience and specialized skills across more than 40 industries and all business functions underpinned by the world s largest delivery network Accenture works at the intersection of business and technology to help clients improve their performance and create sustainable value for their stakeholders. With approximately 384,000 people serving clients in more than 120 countries, Accenture drives innovation to improve the way the world works and lives. Visit us at Copyright 2016 Accenture All rights reserved. Accenture, its logo, and High Performance Delivered are trademarks of Accenture