There is a synergistic relationship between business, analytics, and technology (the BAT triangle) that applies to all business, but is particularly powerful in the case of a property-casualty insurance business. This note reviews the functions and concerns of business: profitability, risk and growth, and explains how business naturally requires analytics for planning, product development and product delivery. Technology provides the tools to implement and communicate results between analytics, business management and customers.
Today accelerating technological development is enabling analysis and disruptive innovation that was not possible even a few years ago. New capabilities are having a profound impact on industry generally and on the insurance industry in particular—a topic we cover at the end of the course.
The importance of technological knowledge was emphasized by Jeff Immelt, then CEO of General Electric, when he announced in 2016
If you are joining the company in your 20s, unlike when I joined, you’re going to learn to code. It doesn’t matter whether you are in sales, finance or operations. You may not end up being a programmer, but you will know how to code.
Part of his motivation: a desire to mimic Silicon Valley entrepreneurs. Programming languages and analytic techniques are becoming ubiquitous in business and any well-rounded employee should have a facility to communicate seamlessly across the three inter-related domains of business, analytics and technology. RMI3338 is designed with these needs in mind.
RMI 3388, Computer Applications in Insurance, introduces students to the use of technology within an insurance company. It provides hands-on examples of the types of project a new analyst would perform, as well as a high-level description of the evolving impact technology is having on insurance.
This document served as an introduction to a course called Computer Applications in Insurance (RMI3388/RMI688) that I developed and taught at St. John’s University, in the School of Risk Management, Insurance, and Actuarial Science. It was written in 2018.
Business is an agreement now to deliver products or services in the future according to specified terms and conditions. The exchange is generally for money. Customers are often strangers with whom there is no prior relationship, nor is there necessarily an expectation of a future relationship.
A business person is someone making business decisions: decisions to bind their company to the delivery of certain goods and services to customers according to set terms and conditions. These decisions are where business lie and are what separates a business person from an advising professional or analyst. The analyst may provide input to the business decision, but unless they are making the decision they remain removed from the business process. Sometimes an analyst makes business decisions in their own domain, e.g. to invest to develop a new analytic capability.
Businesses have three overriding concerns.
Profit can be measured in absolute terms in dollars, as a percentage of revenue or a margin, and as a percentage of resources or capital required by the firm as a return on assets or, more commonly, return on equity. In insurance profitability is often measured using the loss ratio, expense ratio, and combined ratio.
Risk is measured using a variety of metrics including volatility of cash flow and net income, deviation from plan, consistency of earnings, and the ability to meet or exceed target earnings. In insurance, where potential volatility may exceed actual volatility, models are often used to assess a range of potential outcomes. Regulatorsating agencies use these models in order to determine the adequacy of capital.
Growth is usually measured through revenue growth and net income growth. From a shareholder perspective EPS, or earnings-per-share, growth is also used. Ensuring continued growth in the long-term depends on a business’s strategy. Short-term competitive threats from inside the industry must be monitored along with the potential for longer-term disruptive innovations.
A business can be characterized as a series of cash flows with different growth, consistency and absolute value characteristics. Larger enterprises typically combine many different businesses with different cash flow characteristics into a single holding company structure. The holding company’s business is the task of financing its positions in its different businesses as economically as possible using a combination of debt, equity, and other financial instruments. This important business function is usually considered distinct from whatever is the day-to-day “business” of the firm.
Business analytics is an iterative exploration of data, with an emphasis on statistical analysis, used by companies committed to data-driven decision-making. It borrows from, applies and subsumes a wide range of techniques including:
There is an inherent risk in all business operations because businesses commit now to deliver goods or services in the future. The business must plan, forecast and extrapolate potential demand for its products, assess an evolving competitive market, determine optimal delivery, marketing and sales strategies, and invest and deploy resources in advance of revenue to achieve its objectives. These activities must be based on a rigorous analytic foundation, particularly if the business must to raise external debt or equity financing. As a result, businesses are increasingly reliant on analytics for
These services are deployed in product design, pricing and management; process management; risk analysis; marketing sales and distribution strategy and implementation; as well as strategic management and competitor analysis. Keynes’ “Animal Spirits” play a role in new investment, but they are increasingly supported by rigorous analysis.
In many cases analytics is the business, or is a major component of the business. Google, Amazon and Facebook are perhaps the best-known examples of an analytically driven companies. In financial services analytics have overtaken individual human judgments on many critical decision paths.
Bill Gates famously said “We always overestimate the change that will occur in the next two years and underestimate the change that will occur in the next ten. Don’t let yourself be lulled into inaction.” He called for a personal computer on every desk and in every home, but did not dream of an even more powerful computer in every pocket. Soon the entire globe will is connected through the smart phone network, those without electricity using solar panels to recharge their phones.
Since 2012 there has been an unexpected explosion of activity in machine learning or artificial intelligence. Progress has been facilitated by vastly improved computer hardware, including the use of graphical processing units (GPUs), combined with access to truly massive data sets to train the models. Today technology is evolving material capabilities in quintessentially human skills such as
New comprehension, perception and prediction based technologies are being applied in many industries, including financial services and insurance. Underwriting for personal lines and small commercial accounts is increasingly automated, relying on behaviorally-based big data sources to estimate an insured’s underlying loss propensity. As the technology becomes more mature, the hardware cheaper, and the data sources more extensive these trends can only continue in the future. Today, insurance executives need familiarity with key concepts in analytics, technology, machine learning and big data in order to chart a sound strategic course for their firms.
Technology consists of hardware and software.
Many of the successes of computers result from plummeting hardware costs, costs driven down by economies of scale in a virtuous higher-production, lower-cost cycle. It is estimated there are 100,000 mainframe computers, 1.5 billion personal computers, and over 5 billion smart phones. More Indonesians have a cell phone that have access to reliable electricity. A recent iPhone launch sold 25 times more CPU transistors that existed in the world in 1995. As a result of scale hardware has become a low-cost commodity, with cloud providers taking over functions historically managed by a business themselves but with radically lower costs, faster response times, and almost infinite scalability.
Programming creates software that makes hardware come to life. Programming languages vary according to the demands they place on the user, their flexibility and the control of hardware they offer. Lower level languages have high demands of users but allow a lot of control and can be used to create very tailored solutions. Higher level languages are often domain specific and highly optimized but sacrifice flexibility for speed and ease of use. Examples of lower and higher level languages include
The creation of software has been simplified and made more efficient by access to vast software libraries, often at no cost, simplifying the creation of new programs. Google has billions lines of code internally, in a carefully curated and indexed repository; coding often becomes a matter of finding pre-existing solutions and gluing them together with simple scripts. As a result programmers today can be orders of magnitude more efficient than programmers ten and twenty years ago. A great example of technology building under this new paradigm is WhatsApp. Incorporated in 2009, by 2013 it was serving 200 million active users with only 50 staff members.
Technology—the combination of various forms of hardware and software—enables analytics: data gathering and storage; algorithmic computation; distribution and communication of results. Broadly technology can perform a wide range of functions, which can be grouped as follows.
Evolving technology has disrupted many industries. According to Clayton Christensen’s classic model the process starts with a seemingly inferior product that is ignored by incumbents. They focus on building feature-rich products for their most demanding customers. These products over service mid- and lower-tier customers. Incumbents fail to recognize the degree to which the innovation will improve over time and are not prepared when they improve enough to threaten their top-tier customers. Digital cameras are a classic technology-led disruption. Today there are hundreds of InsureTech startup companies trying to disrupt insurance. Potential disruption of the insurance industry could come from innovations based on many technologies:
Insurers are acutely aware of the strategic threats these innovations pose to their businesses. Many are responding by investing in start-up companies in order to be on the inside of future innovations. Designing and implementing an appropriate strategic response to technology-led disruption is high on the agenda of senior managers throughout the insurance industry.
Technology has a dark side, attested by recent election tampering, Equifax data breach and cyber attacks like the WannaCry ransomware. For the insurance industry cyber represents both a threat to operations and an opportunity to grow and sell more. Cyber risk is high profile, and a growing list of senior management casualties has created an urgent senior management-led demand for cyber liability insurance. The insurance industry is struggling to develop the required analytics to quantify aggregation risk and determine suitable pricing for cyber policies. However, even absent robust analytics, the cyber risk market is growing very quickly and is projected to be a $10-20 billion insurance line by the end of the decade.
Business, analytics, and technology (BAT) lie in a symbiotic and mutually reinforcing triangle, Figure \(\ref{battri}\). The power of analytics is deployed to solve a variety of business problems. Low-cost hardware allows efficient and distributed implementation of sophisticated analytic models that can, in turn, be communicated to the businesses and to their customers. As customers use and interact with the solutions the business gathers more data, which can be used to enhance and fine-tune the analytic models, and so the circle continues. Ideally, here is how the BAT symbiotic triangle works:
Despite the obvious synergies, linkages, and relationships between the three corners of the BAT triangle, internally companies often struggle with hierarchical organizations and function-based silos. Each of the three functions tends to attract employees with different skill sets, ambitions, and modes of working. As a result there can be inefficiencies at the interface between the business and analytics services, analytics and technology, and technology in the businesses.
Silicon Valley entrepreneurs speak to the power of sitting at the intersection of the BAT triangle. These individuals usually have a keen appreciation of all three dimensions of their nascent businesses. In too many legacy businesses, however, the relationships are more muddled and the BAT triangle degenerates:
Fundamentally these are problems of education and awareness, which RMI3388 will help address.
The insurance business can be divided into three segments
RMI 3388 will illustrate computer and technology applications within the property-casualty insurance business.
The US property-casualty industry wrote $642 billion of gross written premium in 2017, and, after reinsurance, net earned premium was $546 billion. Business is split almost evenly between personal lines and commercial lines. Personal lines are dominated by private passenger automobile, accounting for 40% of overall premium. Homeowners accounts for a further 6% of premium. The largest commercial lines are liability, including general liability and specialty liability, workers compensation, commercial multi-peril, fire and allied lines, and commercial automobile.
In 2017 the industry combined ratio was 104%, up three points over 2016 because of higher catastrophe losses. The 104% is comprised of a 76% loss and loss adjustment expense ratio and a 28% expense ratio. The industry return on invested assets was 3%, it earned $50 billion of net income, and its return on equity was 5.5%. Statutory insurance companies paid $30 billion in dividends to their stockholders. These statistics are all available on the snl.com website, which you have access to as a St. John’s student.
The risk of the insurance industry was supported by $767 billion of capital, part of $1.9 trillion of assets. The largest single liability was loss reserves of $647 billion.
In order to understand what services the property-casualty industry provides it is instructive to look at the distribution of their expenses. Total expenses, including an average cost of capital, are nearly 50% of premium, corresponding to a spend of over $250 billion per year.
Underwriting expenses can be broken into customer acquisition and management costs, such as sales, acquisition, and needs analysis, marketing, and education, totaling 20% of premium, or $100 billion; and costs associated with providing regulated insurance paper, such as underwriting, product management, regulatory and compliance costs, taxes licenses and fees, billing, policy maintenance and policy issuance, of nearly 10% of premium, or $50 billion per year.
Activity | Expenses | Activities |
---|---|---|
Customer | 100B USD | Marketing, sales, education |
Paper | 59B USD | Maintenance of regulated insurance paper, product design and pricing, customer servicing |
Claims | 69B USD | Claims adjustment services |
Capital | 45B USD | Cost of capital required for risk bearing |
Total | 273B USD |
Table \(\ref{activities}\) summarizes how insurance company operations can be divided into the four categories related to customer, paper, claims and capital.
Insurance companies have become increasingly reliant on technology in all areas of their operations. As a result today it is fair to say that each of the four activities, customer, paper, claims and capital rely on all of the major functions of technology we have identified: databases, easy and hard computations, and communications. Technology is used everywhere.
Activity | Database | Computation | Communication |
---|---|---|---|
Customer | Customer and policy records | Purchase propensity and needs prediction | Marketing |
Paper | Regulatory compliance, underwriting | Credit scoring, catastrophe modeling | Financial reporting |
Claims | Claims process management | Fraud detection | Claim reporting and adjusting |
Capital | Portfolio management | Investment portfolio optimization | Investor relations |
RMI3388 is focused on the insurance business. The course has four basic modules.
These modules track with an insurance business’s focus on profit, risk, and growth, as well as the need to communicate effectively.
The description of RMI3388 says it will
Provide students with hands-on experience in different computer software to perform various data analysis tasks that are commonly required of entry-level jobs in insurance industry. Basic and intermediate statistics concepts are reviewed in the context of insurance applications.
RMI3388 is a multi-disciplinary course, combing technology skills with statistics, accounting, actuarial, underwriting and insurance company operational analytic skills. As a result, the course’s learning objectives can be technology-based, or statistics-, actuarial- or model-based, or both. For example, to quantify profitability the course introduce loss ratios, earned premium and loss development. The relevant actuarial concepts, such as the construction of link ratios, will be introduced through a hands-on spreadsheet implementation. It is not productive to try to separate technology learning objectives from model or statistical ones, and no attempt to do so is made below.
The analysis of loss ratios, loss development triangles and link ratios will be performed in a spreadsheet. We will use data from each student’s selected company and from the CAS Loss Reserve Database.
The Communications section will use a series of case studies based on real-world insurance related and other financial and demographic data. The objective of each study will be to find and communicate a message in the data. Some case studies will be worked interactively in class and others will be assigned as homework. Each student will present one finished exhibit to class.
Case studies will be analyzed using the many different tabular and graphical formats available in Excel.
vlookup
, match
and sortingCatastrophe risk management is an essential part of insurance company operations. It pervades all aspects of the company’s management and is scrutinized by internal and external stake holders. In this module we focus on how hurricane risk is assessed and managed in an insurance company. The module will introduce students to modern catastrophe models. Students will become familiar with the standard modeling terminology, and be able to perform simple calculations given model output.
The idea of simulation will be introduced using a simple model of the 2016 Presidential Election, translated into insurance terms. It will then be extended to a realistic model of US Hurricane exposure. The model will be tested against historical experience.
The strategy and growth module has two parts:
The Bitcoin/blockchain module will try to demystify both concepts. We will introduce hash functions (including discussion and analysis of collision probabilities motivated by the birthday problem) and public key encryption and describe how documents can be digitally signed. Potential applications to insurance, e.g. title insurance, will be discussed. Real-world computations will be performed using a spreadsheet.
We will investigate the Bitcoin block chain and identify specific transactions.
The final course assignment is a group paper and presentation on InsureTech. Students will summarize current InsureTech startup activity. They will then create and administer a survey of friends and peers to identify weaknesses and opportunities in the design and delivery of insurance produces. Based on survey results each group will focus on a particular aspect of disruption, providing either their own ideas or summarizing the approaches of 3 to 5 new startup companies. Each group will write a short paper and present their findings during the final session to the entire class.
Each student will select a US-traded public property casualty insurance company at the start of the course. They will use public information about their company, from 10K and 10Q reports, investor presentations and financial supplements, to illustrate the concepts we study.
Business Problem | Company Illustration |
---|---|
Determine profitability | Reported loss development triangles, including supplements, reported case and bulk reserves, loss ratios by line and in the aggregate, combine ratios |
Communication | Investor presentations |
Measure risk | 10K risk disclosure and catastrophe risk disclosure, including supplements |
Strategy and growth | 10K risk disclosure, investor presentations, research into insured tech investments |
Students are expected to understand the basic operation of computers (e.g. from CIS1332). They should be able to
To the extent students are not familiar with these items we will review quickly in class.