How Technology Can Make Insurance More Inclusive FINANCE, COMPETITIVENESS & INNOVATION GLOBAL PRACTICE Fintech Note | No. 2 © 2018 The World Bank Group 1818 H Street NW Washington, DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org All rights reserved. This volume is a product of the staff of the World Bank Group. The World Bank Group refers to the member institutions of the World Bank Group: The World Bank (International Bank for Reconstruction and Development); International Finance Corporation (IFC); and Multilateral Investment Guarantee Agency (MIGA), which are separate and distinct legal entities each organized under its respective Articles of Agreement. We encourage use for educational and non-commercial purposes. The findings, interpretations, and conclusions expressed in this volume do not necessarily reflect the views of the Directors or Executive Directors of the respective institutions of the World Bank Group or the governments they represent. The World Bank Group does not guarantee the accuracy of the data included in this work. Rights and Permissions The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The World Bank encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. All queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2422; e-mail: pubrights@worldbank.org. Photo Credit: Shutterstock Table of Contents Acknowledgements III Introduction V 1. The Potential of Technology in Insurance 1 2. Why So Many People are Currently Not Served by Insurance 3 3. InsureTech Overview 7 Section 3.1 – Peer-to-Peer insurance 8 Section 3.2 – Blockchain technology 11 Section 3.3 – Concierge Distribution and robo-advice 13 Section 3.4 – Insurance on Demand 15 Section 3.5 – Machine learning, artificial intelligence and big data 17 Section 3.6 – Wholehearted digitization, and the “digitally born insurer” 21 4. Risks 25 5. Recommendations 29 Endnotes 43 TABLE OF CONTENTS I II HOW TECHNOLOGY CAN MAKE INSURANCE MORE INCLUSIVE What are the different characteristics Acknowledgements of DLT? This Fintech Note is authored by Peter Wrede, Senior Financial Sector Specialist in the Finance, Competitiveness & Innovation Global Practice. It is based on many conversations the author had with practitioners of inclusive insurance, with supervisors of insurance, and with InsureTech entrepreneurs, and the author is grateful for their contributions. They are too many to name individually here (but they’ll know they are meant when they read this). The author would also like to express his sincere appreciation to the people who made this report possible, and whose feedback was crucial to the quality of what you are about to read: my thanks to my colleagues Fiona Stewart, Margaret Miller, Susan Holliday, Harish Natarajan, and Douglas Pearce. Appreciation is also extended to Aichin Lim Jones for design and layout support. ACKNOWLEDGEMENTS III IV HOW TECHNOLOGY CAN MAKE INSURANCE MORE INCLUSIVE Introduction New technologies have influenced everyone’s lives since the iPhone came to the market in 2007. Much more than a telephone, the smartphone immediately offered the possibility for remote but connected capture and transformation of data. Capture and transformation of data is at the core of insurance, so the growing community of microinsurance practitioners was excited about the possibilities to overcome administrative obstacles that contributed to exclude the most vulnerable populations from functioning insurance markets. These hopes were generally disappointed in the following years. Smartphones were still rare and expensive, mobile network operators were only just starting to widen their footprint, and attempts to creatively harness the new technology were constrained by limited budgets and expertise of projects often run by people with more experience in poverty alleviation than in insurance or IT. A lot has changed since 2007. Smartphones have become affordable for taxi drivers, rural doctors and informal street vendors, and are ubiquitous in developed countries1. There are apps for everything, and creating a new one does not require large budgets. Mobile banking (on smartphones or feature phones) is providing a reliable and cost effective channel to transact payments with 500 million people, most of them excluded from conventional banking; this matters because transacting payments is the other core pillar of insurance. Furthermore, the growing use of smartphones is generating new types and amounts of data even from people who have scant data trails in formal systems of national identification or credit rating schemes. In parallel, tools to make sense from these overwhelming amounts of unstructured “big” data have emerged, for example in the form of machine learning. This is just one field of artificial intelligence, which is increasingly used in insurance. On microinsurance ventures and the general effort to make insurance more inclusive where it is most needed, the impact of the technological changes that are transforming insurance has not yet been very noticeable beyond insurance transacted via mobile network operators. Projects and practitioners of inclusive insurance still find it hard to justify large budgets for innovative technology and cutting edge expertise. But as FinTech reached the insurance industry to give birth to “InsureTech”, it is insurance companies in the most mature markets who have embarked in the journey to harness new technology in order to provide better products, services and experience to more and new clients. The considerable potential of this endeavor attracts growing numbers of startups and venture capital dollar millions. These efforts are not primarily directed INTRODUCTION V towards low income households and businesses in the regulatory challenge by the Access to Insurance emerging countries. But, as will be argued on the Initiative. Various updates of InsureTech market following pages, they hold the potential to overcome development briefings have also been published. In barriers to inclusive insurance when suitably transferred 2016, InsureTech and inclusive insurance existed to emerging countries. in ecosystems that rarely overlapped, and there was scant discussion on how one could benefit the other. This time it will be different. Attempts in the first post- The purpose of this note is to help bridge this gap, iPhone decade to use technology to make insurance by providing a concise overview of the main lines of more inclusive had the vision and objective but lacked InsureTech considered relevant for inclusive insurance funding and expertise. There is no shortage of funding in emerging markets and developing countries, and expertise in InsureTech, but the focus is mostly on suggesting how they can contribute to inclusion there, mature markets and on people not actually excluded and pointing out some of the risk that regulators and from existing markets. Expanding this focus to also supervisors in particular should be aware of. These embrace inclusive insurance in developing countries InsureTech themes are: (i) Peer-to-Peer insurance; can be transformative for financial inclusion and the (ii) Distributed Ledger Technology; (iii) “Concierge reduction of vulnerability required to escape poverty. Distribution”; (iv) Insurance on Demand; (v) Machine The conception of this note precedes – but complements learning, artificial intelligence and big data; and (vi) – documents and events that have since contributed to “Wholehearted Digitization”. highlight the relationship between inclusive insurance A concise overview of a wide and diverse topic and new uses of technology, such as the International cannot also provide in-depth exploration. That Association of Insurance Supervisors’ Report on contributes to explain why this note continues to be FinTech Developments in the Insurance Industry accurate, but it also means that in several aspects it and upcoming application paper on the use of digital can only provide a high level overview. More detailed technology in inclusive insurance, various Public discussions of particular topics will be warranted in Private Dialogues for Inclusive Insurance organized by the future. Recent developments suggest that client the Mutual Exchange Forum on Inclusive Insurance, data protection is prominent among them. or the Consultative Forum on InsureTech rising to VI HOW TECHNOLOGY CAN MAKE INSURANCE MORE INCLUSIVE 1. The Potential of Technology in Insurance Insurance has long been vulnerable to glitches in manual paper-based processes. In July 2009, torrential monsoon rains flooded the premises of a printing business in Karachi that produced the enrolment forms for health microinsurance that provided cover to tens of thousands of low income households in Pakistan’s remote north. Not being able to print the forms in time for the enrolment window, which was determined by agriculture driven cash cycles, led to a serious setback of the program’s scale and sustainability. Some months later, a forgotten box of claims forms was found at the program’s health care providers, distressing the financial forecasts of various players along the value chain. Some years earlier, the destruction of the World Trade Center in New York had led to litigation, still ongoing in 2009, to determine whether the two planes constituted one or two insured events. This uncertainty, which implied 3.5 billion USD difference in insurance payments, was caused in part by the various fax transmissions of policy forms shortly before the attacks. These are just some examples that highlight the vulnerabilities of paper-based processes in insurance. The intangible nature of insurance suggests that it should be thoroughly digitizable, but that potential is rarely achieved today. The processes underlying the provision of insurance center on the transfer of information and money. Information about the risks to be insured and the conditions for their insurance are exchanged between the insurance company and its client, and eventually constitute the legal basis for the parties’ rights and obligations, which notably include payment of premium and claims. But despite the potential for digitization, insurance worldwide is still dominated by paper based processes. This is because insurance is a relatively old industry, with the largest insurers dating back to the 19th century. Also much younger and smaller insurers have legacy systems and procedures that caution against radical change, and which by and large work to serve existing markets. It is difficult to imagine the provision of insurance without computers today, but their more or less extensive use coexists with manual processes based on paper. The cost this implies puts a lower limit on the size of an insurance policy that can be offered sustainably, and requires patience from customers. Apart from the early implementation of computers which now constitute legacy systems that are not easy to replace, other reasons for the limited digitization of insurance include regulatory constraints, cybersecurity and data protection, and the acceptance by some customer groups which might even face digital exclusion if paper based processes are replaced by digital ones too quickly. 1. THE POTENTIAL OF TECHNOLOGY IN INSURANCE 1 The insurance industry has variously embraced industry6. This enthusiasm takes some momentum technological innovation, but the impact on the from the larger “FinTech” movement, to which The way it interacts with customers has been limited. Economist dedicated a special report in May 2015. For example, leading reinsurance companies in the There is considerable expectation that FinTech will early 90s embraced the tools of artificial intelligence not only disrupt traditional business models but also available at the time to create automatic underwriting contribute to financial inclusion; as Queen Maxima, the systems for life and disability insurance. Assessing the United Nations Secretary General’s Special Advocate insurability of persons with adverse medical history or for Inclusive Finance for Development, put it certain pastimes is a task done by highly specialized underwriters, and their scarcity has been a limiting “The rise of fintech is revolutionizing finance factor for wider outreach of insurance especially in and creating significant potential to spur developing markets. “Expert System” software allowed financial inclusion. Fintech innovators can to codify the heuristics used by these specialists into increase efficiency by making products and decision making algorithms, which could furthermore business processes cheaper, better, and faster. be deployed to the point of sale, shortcutting the They can promote more customer-centric traditional exchange of special questionnaires for financial products. They can really improve special conditions and allowing also high sums to be the customer experience. They can link finance insured on the spot well before the omnipresence of with other development areas such as access to the internet. This technology was crucial for the spread water, electricity, and health care. Finally, they of enhanced annuity products2 but remained limited can improve the cost, transparency, and time to niches. During the dot-com bubble of the late 90s, lags of financial supervision and regulation.”7 several insurance related ventures attempted primarily In fact, many InsureTech ventures, even though to bring insurance advice and distribution online. Some focused primarily on developed markets so far, could online comparison sites evolved from there, but few be considered to pursue an inclusion agenda, as they ventures survived when access to capital dried up. In aim to attract people and businesses to insurance who, 2013, research by the Boston Consulting Group and while not outright excluded, have not been attracted to Morgan Stanley showed that consumer satisfaction with traditional insurance markets and models. online experience of insurance lagged considerably behind that of most other industries3. Until recently, The importance of a more comprehensive technological advances have not questioned the core digitization of insurance business depends on the business model of how insurers serve their clients. But circumstances. Despite increasing pledges of customer Inga Beale, CEO of Lloyds of London, writes: centricity, insurers in mature markets can continue to serve their traditional client groups as they have done “It is no secret that as an industry we are in the past. This also includes the limited segments lagging behind the rest of financial services of society that insurers mostly serve in developing when it comes to digitalization and the use of countries, people who also have bank accounts, mail new technologies. There are no more excuses. addresses, and familiarity with filling forms. Their If we don’t adopt and embrace new technology, discontent with current insurance practices is moderate, we won’t have a future.”4 and so is the motivation to invest heavily in technology “InsureTech” denominates a recent trend that aims and risk the uncertain outcomes of possible disruption. to revolutionize how insurance is done. Starting to get Reaching other customer groups, however, will require noticed in 2015 – that year’s InsureTech startup funding a different approach. In developed markets, such groups amounted to 2.65 billion USD – statements like “The include millennials and adherents of the “Sharing insurance industry is on the brink of major technology- Economy”. But in developing markets they include driven change”5 and “insurance will change more in the many more potential customers, such as the poor and next 5 years than in the last 100 years” are becoming the wide majority of the population not fully served by more frequent. More than 1,300 global startups are insurance so far: (emerging) middle classes, non-poor focusing on technology applications to the insurance rural populations, and women. 2 HOW TECHNOLOGY CAN MAKE INSURANCE MORE INCLUSIVE 2. Why So Many People Are Currently Not Served By Insurance Most people on the planet are excluded from insurance markets today. Although the (re)insurance sector in OECD countries alone has about $23 trillion of assets under management (that’s about a quarter of global GDP), insurance penetration8 varies widely across countries, from 36% (Cayman Islands) to 0.04% (Guinea)9. But as insurance penetration often reflects primarily commercial, industrial, engineering and mining activities in developing countries, this indicator sheds little light on the use of insurance by households, and it is difficult to quantify how many people currently have access to, and use, insurance. That is because insurance companies report premium volume to investors, supervisors and the public, but not usually the number of insurance policies they have in force or underwritten during a period, much less the number of unique individuals who hold these policies (this requires consolidating multiple policies). The few examples who appear to do that include the IMF’s publication of the number of “policy holders with insurance corporations” in 2015 for 22 countries10, showing that 10% of people in Bangladesh have insurance, as do 4% of people in Rwanda, 1% of people in Mongolia, and 0.7% of people in Guinea. The reasons why insurance is not used by more people are well understood. Considerable research has been conducted to understand the low usage of formal insurance, especially among vulnerable populations who are expected to benefit most from suitable risk management in their struggle out of poverty. While local particularities play a role, there are a number of common factors that explain low insurance uptake. They need to be understood to appreciate how suitable use of technology like that proposed by InsureTech ventures has the potential to overcome these barriers. Barriers to inclusive insurance (1): limited purchasing power. In absolute terms, people with low income can spend less on insurance. They need lower insurance cover, as the overall value of their assets is lower, as is the loss of income due to death or disability. But the scalability of insurance premium is limited: while lower sums insured translate into proportionally lower actuarial risk premiums, the margins insurers need to add for administration, distribution and other expenses are not perfectly scalable, as there is a minimum cost incurred in the administration and distribution of any insurance policy. Barriers to inclusive insurance (2): limited understanding. Numerous surveys in developing countries show that most people there never had any insurance, never 2. WHY SO MANY PEOPLE ARE CURRENTLY NOT SERVED BY INSURANCE 3 heard of it, and misunderstand it. Despite increasing does not work for low income people, whose insurance efforts towards financial literacy of wider populations, needs cannot be met with products inspired by the US that is not changing quickly. The focus of financial or Germany nor with local products “downsized” to a inclusion strategies and financial literacy efforts often lower premium. Products for inclusive insurance need is primarily on banking products – loans, savings, to be developed with good understanding of the target payment systems – which are more immediately population’s circumstances and needs as well as with relevant to most people, easier to understand, and insurance and actuarial expertise. When that is not the easier to provide. They are comparable to the informal case, microinsurance products are not met with demand. mechanisms that most unbanked people use to save Barriers to inclusive insurance (5): unsuitable or borrow. By contrast, insurance is a much more distribution. Of all the aspect of insurance provision abstract concept which doesn’t immediately relate to that need to be reengineered to make insurance work the various informal risk sharing mechanisms in use by for low income people, distribution is the most critical. low income communities. On the one hand, the cost per policy of traditional Barriers to inclusive insurance (3): limited trust. distribution – notably agents and brokers – makes it Even if they have no direct or personal experience with unsustainable for policies with low premiums. Unless insurance, many people view it negatively. Publicly the commission makes up a disproportionate share of vented anger of policyholders dissatisfied with their the premium, these channels will not be able to serve insurer’s claims service – often in relation to mandatory emerging customers under this model of distribution, insurances, e.g. for cars or motorcycles – is difficult to as their efforts cannot be remunerated appropriately. judge for anyone with limited understanding of the On the other hand, upfront investment cost (in product basics of insurance. In countries with former state development, marketing material etc.) needs to be monopoly insurers, years of poor service have often recouped from a very large number of policies when tarnished the reputation of insurance so much that even each has a low premium, making microinsurance after de-monopolization, markets have difficulty in an example of a high-volume-low-margin business rebuilding trust. Low income households are frequent model. But to achieve the high volume, very effective victims of financial scams that exploit their limited distribution is required. That usually requires partnering financial literacy, so they are understandably wary. with third parties, and the management of this longer Poorly adapted processes to provide insurance to low value chain can easily break down. income households often further add to disappointment Barriers to inclusive insurance (6): unsuitable and distrust, for example when insurance agents give business models. Most insurance companies in up on rural communities and stop coming to collect developing markets focus on insuring large commercial premiums (resulting in involuntary policy cancellation) accounts and mandatory motor insurance. They are or when alternative distribution channels such as not experienced in selling to, and serving, very large mobile network operators fail to explain product numbers of individual policyholders, and are not features thoroughly. Lastly, the conventional approach experienced in administering millions of policies. Nor to insurance provision does not seek to interact with are they experienced in disaggregating the value chain clients regularly; instead, they are contacted once and partnering with other players which can perform a year to remind them of premium payment. This specific tasks better and more cheaply, for example underdeveloped post-sales culture does not contribute explaining products, doing the enrolment paperwork, to nurture the confidence of skeptical first time buyers. and collecting premium. Their processes are based Barriers to inclusive insurance (4): unsuitable on abundant use of people and paper, as pressure to products. Given the lack of actuarial resources and increase cost efficiency and speed are low in markets statistics, insurance products in developing countries where insurers still can grow nicely serving the upper often replicate products in developed markets classes only. (disseminated by reinsurers). That may work for upper To a surprising extent, InsureTech ventures in socioeconomic groups, people with cars, urban houses developed markets are driven by similar barriers. with property titles, and formal employment. But it Even in the US and Europe, the use of insurance remains 4 HOW TECHNOLOGY CAN MAKE INSURANCE MORE INCLUSIVE behind its potential, because large numbers of people excluded from formal insurance markets but focus on distrust conventional insurance business models and people who voluntarily remain underinsured in the don’t find their needs reflected by existing products and existing models of insurance provision. But if they procedures. Most InsureTech propositions don’t aim succeed at that, they also have to potential to overcome to make insurance work for people who are currently barriers to inclusive insurance. The G20 High Level Principles for Digital Financial Inclusion • PRINCIPLE 1: Promote a Digital Approach to Financial Inclusion • PRINCIPLE 2: Balance Innovation and Risk to Achieve Digital Financial Inclusion • PRINCIPLE 3: Provide an Enabling and Proportionate Legal and Regulatory Framework for Digital Financial Inclusion • PRINCIPLE 4: Expand the Digital Financial Services Infrastructure Ecosystem • PRINCIPLE 5: Establish Responsible Digital Financial Practices to Protect Consumers • PRINCIPLE 6: Strengthen Digital and Financial Literacy and Awareness • PRINCIPLE 7: Facilitate Customer Identification for Digital Financial Services • PRINCIPLE 8: Track Digital Financial Inclusion Progress Numbers 2, 3, 5 and 6 seem particularly relevant and applicable to inclusive InsureTech, and provide guidance both to national and international discussions and knowledge sharing. 2. WHY SO MANY PEOPLE ARE CURRENTLY NOT SERVED BY INSURANCE 5 6 HOW TECHNOLOGY CAN MAKE INSURANCE MORE INCLUSIVE 3. InsureTech Overview What’s really new about InsureTech is the branding, and the critical mass it attracts. According to FinTech Weekly, “InsureTech or Insurance Tech are technologies and platforms that help optimize any of the principles for success or requirements of insurance. By extension: any company that provides insurance though the engagement of technology in a user-centric way.” Insurers have employed technology in the past, but what is different now is that previously unseen amounts of entrepreneurial creativity turn towards insurance and are met with previously unseen amounts of investor funding. Supporting factors are new insights in behavior, and recently maturing technologies and digital ecosystems such as artificial intelligence, distributed ledgers, or the analysis of big quantities of data, as well as demographic changes and faster progress towards customer-centric technology in other industries. After years of low IT investment, the insurance industry acknowledges a pent up need for new capabilities and technologies. Consulting companies and organizers of conferences further fuel the sense of paradigm shift, and supervisors are starting to address new questions systematically. InsureTech ventures can be clustered around a few major themes, not all of them equally relevant for inclusive insurance. The themes that this report will focus on are: • Peer-to-Peer insurance (P2P) • Blockchain technologies • “Concierge Distribution” • Insurance on Demand • Machine learning, artificial intelligence and big data • Wholehearted digitization The terminology is not standardized yet: in the media, the concepts are grouped in different ways (for example by including P2P under Insurance on Demand sometimes) and called different names (for example Distributed Ledger Technology instead of Blockchain, or Robo-Advice instead of (certain aspects of) Concierge Distribution). Themes that will not be discussed here include the Internet of Things (including wearable devices and telematics) because they still rely on devices that low income populations in developing countries will find too costly, and the various ventures 3. INSURETECH OVERVIEW 7 that aim to help consumers navigate the complex will submit fraudulent claims. Insurance is perceived landscapes of health insurance in the US and Canada. as a zero-sum game in this mindset, and antagonism Neither will possible changes in insurance needs related seems inevitable. P2P restructures risk pooling such to emerging technologies such as autonomous vehicles that claimants know who will sacrifice income to nor the increasing importance of cyberinsurance be compensate their loss, or such that they know that it is discussed, assuming that their contribution to make not the organizer of the insurance who will benefit when insurance more inclusive in developing markets will they don’t submit a claim. likely be limited in the near to mid future. P2P hopes to reconcile the law of large numbers with the “law of small numbers” which states that Section 3.1 – Peer-to-Peer insurance (P2P) trust does not extend beyond a limited number of P2P harnesses technology to connect the insurance persons. The challenge of all attempts to base reliable experience with its roots in organized mutual insurance on existing trust networks (for example in solidarity. Besides using savings, the mechanism most the case of village based health mutuals) has been that widely used by low income households in developing the size of the initial solidarity group was too small to countries to cope with unexpected shocks is help from bear more than minor events, and that the law of large friends and family. While this help is often provided numbers that allows to statistically predict claims in in a spontaneous way, more formalized arrangements insurance did not apply. This put pressure on these pools of mutual help have evolved in many places, such as to scale up and include more members, but in doing so funeral societies, local health mutuals, and village the original trust eroded and the inclination to overuse based organizations. They all constitute organized the cover increased, often to the point of rendering forms of reciprocity that are more predictably reliable, the scheme inviable. Another challenge of such small and they are usually based on social capital and mutual schemes has been that in addition to trust, expertise is trust of its members. Starting in 2010, German startup required to run them effectively, but their small size Friendsurance began replicating this concept replacing makes the required resources unviable. P2P aims to village or neighborhood communities with Facebook overcome that with greater use of new technology like groups of friends who committed to help indemnify social media, which have made communities wider and any member whose bicycle got stolen. This initial proof more diversified for many people today. of concept evolved into partnerships with insurers who Insurance fraud adds to the cost of insurance, and by provide policies with higher than usual deductibles and reducing the amount of fraudulent insurance claims mutual solidarity groups (organized by Friendsurance) with better alignment of interest, P2P suggests who help their members pay the deductible in case lower premiums without compromising on cover. of claim. Fueled by the success and media coverage The Coalition Against Insurance Fraud estimates that in of Peer-to-Peer lending in recent years, various other the US alone, insurance fraud amounts to at least $80 startups are testing various models of P2P insurance. billion a year across all lines of business. Fraudulently P2P aims to bring trust back into insurance. Dan staged car crashes cost UK insurers an estimated 340 Ariely, leading researcher and author of books on million £ per year11. Inspool, a P2P startup aspiring to behavioral science, says “unite and reward polite drivers”, claims that 15-20% of conventional car insurance premium covers the cost of “If you try to create a system to bring out the “boy racers”, that 5–15% account for false claims, and that worst in people, you’d end up with one that “95% of people end up paying for the 5% of hooligans”12 looks a lot like the current insurance industry.” . Friendsurance’s model to bring high deductibles into He says this in his function of Chief Behavioral Officer of otherwise conventional insurance policies serves the Lemonade, a startup with initial P2P appeal, to emphasize same purpose to reduce the incentive of fraudulent the perceived conflict of interest between insurers and claims – often smaller claims that fraudsters hope will not insureds that has so widely eroded trust between them. merit thorough investigation – and this is why German Many customers believe that insurers will do anything insurers like the approach despite the lower premium to deny a claim, and insurers worry that many customers they receive due to high deductibles. And even when not 8 HOW TECHNOLOGY CAN MAKE INSURANCE MORE INCLUSIVE fraudulent, small claims contribute disproportionately to “In many cases, mutual insurers were originally an insurer’s expense ratio, so encouraging people not to set up by specific socio-economic groups (such as claim for damages that they can absorb well themselves farmers, fishermen and teachers) in the absence – e.g. by peer pressure in P2P models – contributes to of suitable protection or savings solutions make insurance more cost efficient. from the mainstream insurance sector. Where there is a great deal of ambiguity about the Transparency is another distinguishing feature distribution of possible insured losses, risks may of P2P insurance. Much distrust of consumers is become uninsurable for commercial insurers or owed to their perception that insurance premiums are protection might become prohibitively expensive. unnecessarily high because of lavish office buildings Mutuals can often insure their member-owners and excessive executive pay. Inspool’s advertising states at affordable premiums.”13 that insurers’ overheads are 10-25% and their profit margin 10-15%. So a transparent charging structure and highlights the similarities at least of small traditional tends to be at the heart of P2P ventures. Lemonade will mutuals with P2P, for example in respect of members’ take exactly 20% of premium to cover its cost (and rights and stake in generated surplus. However, the thereby risk not being able to recover upfront and other report also remarks that P2P will have to achieve non-volume-proportional cost if the business volume significant scale if the concept is to compete noticeably falls short of expectations). Chinese P2P company with conventional insurance. Widespread adoption of TongJuBao takes 25% of the money pooled. Other blockchain technology (Section 3.2) may help achieve P2Ps who don’t have such a simple charging structure that scale, as this technology is well suited to support go to great lengths to explain pricing in detail. disintermediated financial services such as Bitcoins. P2P also appears to disintermediate insurance. The P2P shares many similarities with Islamic fact that there is an administrator who facilitates the peer Insurance (Takaful). Under the Wakala model of risk pool can be eclipsed by the stronger emphasis on the Takaful, peers form and join a risk sharing arrangement pool members and their role (depending on the model) and delegate the administration to a Wakeel who has to decide which covers to provide and which claims to the specialized expertise to do that efficiently. The pay. Furthermore, conventional insurance distribution Wakeel is paid an administration fee while all surplus by agents or brokers is not seen in P2P insurance which generated by members’ contributions exceeding claims relies on advertising, social media, word of mouth and payments reverts to the members. The Wakeel thereby social groups to spread awareness and attract more pool has no incentive to deny claims, and (in theory) the participants. This disintermediation further contributes expenses associated with running a Takaful scheme to reduce cost. are transparent. One significant difference to P2P is that when the funds collected from members are not Disintermediation allows new responsiveness. The sufficient to pay all legitimate claims, the Wakeel has to greater involvement of risk pool members not only provide an interest-free loan to the pool which is repaid reduces cost, it also allows to provide insurance for from future surpluses. Nevertheless, the general features risks that could not be insured previously under some that distinguish P2P from conventional insurance models. As it is the customers themselves who bear the (transparency, mutual solidarity, and the absence of risk that premium will be insufficient to cover all claims, profits based on others’ misfortune) are likely to make it is primarily their risk appetite and awareness which the concept appealing also to Muslims, and the greater determines what covers are tested, allowing for more cost efficiency aspired by P2P can make it a viable experimentation than with conventional insurers. While model of microinsurance in the Muslim world. mostly applied to less vital covers such as “divorce first aid”, this approach also allows to provide cover for Various different models for P2P are currently serious risks that were previously not insurable. being tested. While most P2P operators are specialist intermediaries (between members, or between P2P has been portrayed as an evolution of mutual members and insurers), Lemonade provides an example insurers. In its Sigma report 4/2016, Swiss Re points of a venture initially (but no longer) branded as P2P out that with insurance license. Its business model differs 3. INSURETECH OVERVIEW 9 considerably from that of other P2P schemes in that risk whole, with detrimental consequences for consumer is not borne by the members and surplus is not returned trust. Where regulation is in place that requires insurance to them. Instead, surplus is donated to nonprofit causes premium and conditions to be submitted for approval of the customer’s choice, so that Lemonade does not by supervisors – for example to assess client value or benefit from low claims. Not the outlook to get money check for possible discrimination of customer groups back appeals to clients, but the promise of low expenses, – it may conflict with P2P’s tailor-made approach to convenient interaction (app and bot based), and the the cost of insurance. The protection of customer data aspiration to provide “insurance as a social good rather also requires particular attention in P2P insurance. The than a necessary evil”. While TongJuBao marks one National Association of Insurance Commissioners in the end of the P2P continuum where no insurance company US thinks that is involved (see Box), Friendsurance – who combine “Although, P2P insurance could and should P2P elements with conventional insurance – and most be regulated like any other insurance company other P2P propositions lie somewhere in between and within the existing regulatory framework of state are often intermediaries rather than underwriters. regulation, this innovative model of managing Some regulators have doubts about P2P. They wonder and delivering insurance products presents a if there can even be P2P insurance, or if both concepts new challenge for state insurance regulators to are mutually exclusive. Are “P2P” models based on a study its strengths and weaknesses as well as its conventional insurance license (like Lemonade’s initial differences from traditional insurers.”14. branding) substantially different from existing insurance, or is the P2P label primarily a marketing aspect? On the The fact that the majority of P2P ventures operate other hand, insurance regulators and supervisors have within existing insurance regulation documents their little sympathy for P2P that is not subject to insurance efforts to comply with the rules but hampers their regulation and supervision, pointing to the obvious ability to implement genuine P2P models that transcend reasons why clients and financial markets need to be conventional insurance. International experience with protected from unbridled entrepreneurship, all the the TongJuBao model will show if P2P can work more so if they are emerging customers and first time outside of insurance regulatory frameworks. buyers of insurance. And even though P2P ventures try P2P holds promises for more inclusive insurance. In to differentiate themselves as much as possible from the developed markets where it is being tested so far, P2P insurance as we know it, a spectacular failure risks models justify their approach with alleged widespread tarnishing the reputation of the insurance industry as a rejection of conventional insurance which leads many A Radical Approach To P2P Insurance – The Example Of TongJuBao Launched in 2015, the Chinese company TongJuBao provides an example of P2P without any insurance company involved, showing the possible degree of disintermediation. Instead of partnering with or intermediating for insurers, TongJuBao is a matchmaker for people willing to join a common risk pool under an arrangement governed solely by civil law contracts. In addition to the matchmaking, TongJuBao also provides administration of the risk pools (using apps and chatbots to facilitate transparency and ease of interaction), but legal entitlement of pool members is only among themselves, not towards TongJuBao. The company believes that this or a similar model will also work outside of China and is preparing to launch in France and the US, where 85% of surveyed persons have reacted very positively to the value proposition. TonJuBao provides a good example of how the absence of insurance formalisms can result in new covers that respond to users’ needs, and has gone further than other P2P ventures to establish mechanisms like discussion boards and voting procedures to implement new cover ideas (even when they are not endorsed by a majority). One example is the child safety cover: it addresses the threat that children are abducted for child trafficking, an event of low likelihood but very high severity. Should that happen to a family, additional resources to employ private investigators in addition to the standard police procedure can greatly increase the probability that the child is recovered, but most families do not have the means for that. This is where TongJuBao’s cover makes the difference, providing money for extra search efforts also to families of modest income. 10 HOW TECHNOLOGY CAN MAKE INSURANCE MORE INCLUSIVE to remain uninsured or underinsured. The reasons spending” challenge. Until then Digital Assets could not for this rejection are similar to demand obstacles for be exchanged in a peer-to-peer manner, as there was no inclusive or micro insurance in developing countries: effective way to ensure that the digital asset is indeed lack of trust in commercial insurers, preoccupation in the possession of “sender” and has not already been about fair prices and conditions, the absence of covers sent across to another person/entity. Blockchain has the that respond to consumers’ needs, and premium that potential to become for assets and transactions what the many deem unaffordable. So far, P2P ventures in internet is to information. developing insurance markets are few, with examples Blockchain technology has the potential to reinvent in Colombia (Wesura) and South Africa (Riovic’s trust. Central to the concept of a distributed tamper- partnership with PeerCover) in addition to China. One proof and fully auditable ledger is the absence of a trusted likely explanation is the uncertainty about what degree intermediary. Instead, parties participating in blockchain of true P2P risk pooling will be acceptable to regulators based transactions (e.g. payment with Bitcoins) trust the and consumers, given that almost all current schemes system. That allows to do transactions with strangers are very young and untested. without risking counterparty default, a risk traditionally Given that mutual solidarity is much better understood mitigated by (central) banks, PayPal, stock exchanges, than formal insurance in many of the microinsurance or insurance companies. It explains why The Economist target populations, portraying conventional insurance calls blockchain technology a “trust machine”16. Not more as P2P (like Lemonade does) could facilitate needing a trusted intermediary reduces the cost of understanding and acceptance of first time buyers. transactions and expands the universe of people you can Rather than implementing novel business models, this transact with, for example in respect of remittances or in requires insurers to make an effort to increase the sense the context of the “sharing economy” (where blockchain of mutuality – possible for example in profit sharing based contracts can formalize lending relationships and arrangements of group life – and the transparency of thus make them accessible to insurance17). But blockchain their operations, and seek ways to be perceived more as technology also offers the potential to reduce operating administrators of risk pools (like the Wakeel in Takaful) costs of transactions that so far require verification, as than as parties who benefit from rejecting claims. currently tested regarding the exchange of data between reinsurance and insurance companies18. Section 3.2 – Blockchain technology (also Blockchain-based trust could support known as Distributed Ledger Technology) disintermediation also in insurance. Insurance Distributed Ledger Technology refers to a novel and companies are an example of trusted intermediaries fast-evolving approach to recording and sharing data that guarantee that the risk pool that I have been across multiple data stores (or ledgers). This technology contributing to will honor my claims when I deserve allows for transactions and data to be recorded, shared, compensation. Where this guarantee is instead based and synchronized across a distributed network of on individual contracts under civil law between all different network participants. A ‘blockchain’ is a members of a risk pool – like with TongJuBao – the particular type of data structure used in some distributed role of the insurance company may seem obsolete, ledgers which stores and transmits data in packages but basing risk pools on individual contracts between called “blocks” that are connected to each other in a all of their members seems cumbersome; and even in digital ‘chain’. Blockchains employ cryptographic and the P2P case there is often an administrator to perform algorithmic methods to record and synchronize data some tasks that require trust, e.g. in relation to claims across a network in an immutable manner15. adjudication. Blockchain suggests that (P2P) risk pools can be conceived without any intermediary (but Blockchain technology enables digital assets. Based possibly with some form of reinsurance), based on a on cryptography work from the 1990s, the concept of more efficient form of contract between all participants. a distributed ledger managed autonomously on a peer to peer computer network was operationalized in 2008 Blockchain technology allows “smart contracts”. and put to practice with the digital currency Bitcoin Contractual relationships documented in computer code in 2009, when the technology overcame the “double can include elements that conditionally auto-execute. 3. INSURETECH OVERVIEW 11 For example, a life insurance contract fully represented arrangements will often not be considered insurance if in a blockchain resident in a computer network linked there is no licensed insurance company involved and to a web-based publication of death certificates could product conditions and rates have not been approved. trigger payment automatically when the death certificate Blockchain-based arrangements involving a licensed of the insured person is uploaded. Or a smart index insurer, on the other hand, are yet to prove their cost/ insurance contract linked to sources of weather data benefit advantage. The legality and enforceability of a such as satellites or automated weather stations could blockchain based contract is unclear at present, as is autonomously trigger payment when the insured index the legal entity corresponding to a blockchain based crosses the defined thresholds, having verified that the arrangement. Hence an administrative step might be policy was in force and that the premium had indeed required to transform a contract based on blockchain been paid. Such elements of self-execution promise code into a legal document in the conventional (further) cost reduction both in conventional insurance sense. Other concerns relate to the location and (various re/insurance companies are already testing security of personal data that often is subject to data blockchain applications, including for microinsurance19) protection rules. And contracts that make claims and in P2P (ventures based on blockchain are emerging). payments automatically raise new questions about It also increases speed of transactions, addressing a solvency requirements appropriate to safeguard the common complaint of insurance customers. A prototype ability to honor all claims at all times, as well as the of blockchain-based insurance for flight delays has been role of human judgement in claims adjudication. developed by InsureETH20 with flight delay information However, some regulators are encouraging the fed directly from the web. assessment of blockchain uses in less disruptive ways. For example, the UK’s Financial Conduct Authority Blockchain also allows conditional use of payments. director of strategy and competition recently told the Remittances and insurance payments in conventional Financial Times they were currency can be used to purchase anything. Payouts from health insurance can be used to pay rent, and “talking to firms thinking about how to apply payouts from agriculture insurance can be used to pay that to financial services and how it could school fees. While in some circumstances that may not benefit consumers or indeed make the business matter, in others it does. The impact on family health, of compliance easier”21 for example, may not develop as anticipated if health Blockchain technology can help overcome several insurance payments are not used towards health care. of the barriers to inclusive insurance discussed in But even when they are, the outcome may still fall short section 2. The reduction of administrative friction and of expectations if the payouts are used for substandard cost hopes to reduce the price of insurance (assuming health care or counterfeit pharmaceuticals. Payouts of the initial investments in blockchain technology are blockchain based insurance can be conditional on the amortized over a large enough number of policies), thus use, for example that health care services are procured alleviating constraints in the ability to pay. Expectations at quality-certified empaneled providers. Conditionality that blockchain can help reduce payment for fraudulent of payments can even go further up the value chain, claims through crowd-sourced verification may further by guaranteeing for example that health care providers reduce the cost of insurance. Increasing the speed procure pharmaceuticals from quality-guaranteed supply of processing – especially for claims – will benefit chains. Furthermore, insurance executed via smart the customer experience. The blockchain premise, contracts could automatically generate statistics on its if properly understood, replaces trust in insurance use – when do most car accidents occur in which part of companies – often challenged - by trust in a system and the city, when do diagnoses and treatment for a particular might eventually contribute to a more rational approach disease peak – that could themselves trigger actions to risk management. related e.g. to redesign of the insurance cover or to reinsurance when combined with suitable data analytics. The exploration of blockchain technology to insurance is nascent, and its impact on inclusive Insurance regulators started discussing blockchain insurance will not be immediate. But this exploration applications. As noted above, pure P2P risk pooling is pointing insurers in mature markets to consumer pain 12 HOW TECHNOLOGY CAN MAKE INSURANCE MORE INCLUSIVE points that are bigger obstacles to insurance inclusion cost of leaving people alone in their purchase decisions. in less developed markets. The transferability of Concierge Distribution and robo-advice harness new the technology (apart from local regulatory aspects) technologies to bring back assistance without the cost it means that substantial innovation, once prototyped and used to carry (although some ventures combine chatbots tested in developed markets, can provide a pathway with experts in callcenters). It also replaces paper- for adoption in developing markets with appropriate based with digital processes, increasing convenience modifications to take into account local context. for customers (accustomed to single sign in of cloud- based applications like the google/android ecosystem) Section 3.3 – Concierge Distribution and and cost efficiency for insurance when systems are set robo-advice (also referred to as platform up to appropriately interface. based models) Concierge Distribution also promises to bring Concierge Distribution is the evolution of price values to insurance that are important to new target comparison websites for insurance. It is provided markets. While most customers like client-centricity, by smartphone applications which not only promise to and convenient hassle-free immediate transactions, help buyers find the best price for the insurance they are millennials in particular have come to expect that from looking for, but also to help them realize what insurance any company they do business with. But insurance they are – or should – be looking for, often involving has been slower than other industries to meet these chatbots which support the customer in this needs expectations: its customers interact only sporadically assessment (hence the term “robo-advice”). Simplified with their insurer, and are less satisfied with their enrolment and underwriting with cooperating insurers interaction experience than customers of banks or may provide increased convenience when the retailers. Consumer satisfaction with online experience information required in insurance forms already resides is particularly low for insurance, half that of airlines for in the app. This allows to streamline what is often example23. At the same time, consumers are increasingly experienced as a cumbersome process especially by accustomed to be in control of their financial and digital natives accustomed to immediate transactions purchase decisions, in contrast to the traditional way of and high convenience. Additional services offered by buying insurance which is often perceived as opaque. some apps include the possibility to import already Being in control includes sharing feedback – negative existing insurance contracts for easy retrieval and if necessary – and thereby pressure service providers administration (for example to optimize or check for to pay due attention to their customers’ experience. duplications of cover), and to receive reminders for Being in control also includes the possibility to easily premium due dates. Some apps even provide support switch service providers, something that has not been when submitting claims, for example via chat and straightforward in conventional insurance. Ultimately, photographs. In health insurance, further service being in control means that consumers meet providers components can include the facilitation of doctors’ at eye level, an experience that has been the exception appointments and prescription management. in insurance because of the asymmetry in understanding of technicalities. Concierge Distribution thus aims to digitize core service components of insurance, to reduce cost (Small) businesses may benefit even more from and improve user experience. Best advice based Concierge Distribution. Managing the insurance on customer needs assessments is a task traditionally needs of enterprises is considerably more complex than performed by insurance brokers or agents who often for households, especially if they operate in the formal have a long term trusted relationship with the client. sector with its various requirements for mandatory Their services are appreciated but come at a cost and insurance. Large companies have specialists in charge usually delay the process of acquiring insurance, and of this task, and may negotiate favorable terms with a majority of customers are no longer willing to pay insurers. Small businesses don’t have that advantage. for face-to-face services22. Disintermediation based SMEs operating in the informal economy may not on online comparison facilities and web-based self- comply with requirements for mandatory insurance but enrolment has reduced the price of insurance but at the are still exposed to a number of personal and property 3. INSURETECH OVERVIEW 13 risks that are seldom well met in developing countries. association with new and un- or differently regulated CoverWallet for example is a US online broker offering players in the value chain. Regulators in countries Concierge Distribution services to (small) businesses, where such “m-insurance” has grown strongly are addressing covers like Workers Compensation and increasingly aware of the need to address this. Directors & Officers liability. Inclusive insurance can benefit in various ways from Concierge Distribution raises regulatory questions, the concept and technology of Concierge Distribution. in particular the robo-advice component. Bringing transparency to insurance products, prices Distribution of insurance is subject to market conduct and conditions can be expected to increase customers’ regulation in most jurisdictions, in particular the act value for money, and improve affordability. Training, of providing sales advice and making binding offers. licensing and professional liability insurance contribute The corresponding regulation usually requires special to make conventional distribution channels too costly qualifications for people who sell insurance, often to service the majority of low income or geographically asking they be licensed. The liability for inappropriate remote populations. Regulatory requirements for advice – which can cost customers dearly when it disclosure (often requiring paper based information) leads to wrong insurance purchase decisions – tends and advice further increment the cost of distribution. to be clearly stipulated. This needs to be clarified for Concierge Distribution can reduce cost in these areas, Concierge Distribution: are chatbots merely providing and with growing prevalence of smartphones can advice, or are they mediating insurance? Other issues to reach populations that other distribution channels consider relate to rules about the protection of personal have failed to serve sustainably. The customer-centric data. It is not obvious who owns the data provided to a aspiration of Concierge Distribution can help overcome Concierge Distribution app, nor where they reside and distrust, when the app and the chatbots or humans how securely they are stored. Furthermore, the claim that provide the advice are perceived to genuinely that a Concierge Distribution app (and the call center advocate for the customer. It can also overcome the staff when part of the service) provides truly independent lack of understanding, when low cost communication advice and facilitates to buy the best possible insurance technology such as chatbots can (repeatedly) explain for a consumer’s needs and circumstances is not easy the basics of insurance in general, and the particularities to verify for the consumer, who knows little about of a person’s cover, whenever they have any doubt and the incentive structures guiding the (human or robo) at typical pain points such as renewal or claims stage. advice. Some InsureTech distribution startups merely This has the potential to make insurance less opaque provide potential customer leads to the insurers they and reduce understanding asymmetries that make first partner with, or they are geared towards upselling, with time buyers uneasy, while replacing the hassle of filling little aspiration to best advice. forms by providing information through interactive chats. Concierge Distribution furthermore not only aims While the transparent comparability of insurance to make buying insurance but also having and using premiums is advantageous for consumers and can insurance more pleasing. It could thus complement be expected to stimulate competition for lower rates freemium and other insurances sold via mobile network and more generous conditions, this competition risks operators in Africa and Asia which struggle with going too far and leading insurers if not to insolvency worryingly low utilization. then to liquidity issues24. And Concierge Distribution inherently aims to commoditize insurance and be The innovative use of smart phones that characterizes the primary brand and interface to (and owner of) Concierge Distribution (for example to make car the customer. While such white-label approach to insurance proposals based on a photograph of the distribution partnerships is not new, tends to be license plate25 is already being tested in other attempts acceptable to insurers when the partner’s distribution to make insurance more inclusive26. potential is stronger than their own – for example m-insurance, that is simple insurance intermediated when microinsurance is provided by mobile network by mobile network operators (MNOs), could be operators-, it exposes the partnering insurer and the considered a precursor of Concierge Distribution. entire insurance industry to reputation risks from their Predating the emergence of InsureTech, this form of 14 HOW TECHNOLOGY CAN MAKE INSURANCE MORE INCLUSIVE insurance is not supported by technology innovations 2013 to 335 billion USD in 2025, matching the size of beyond an (ideally) seamless integration of data the traditional sector.28 administration platforms of the insurer, the MNO IoD also questions why insurance contracts have to and usually also a technology service provider, and provide all-inclusive cover. Insurance against theft or convenient payment of premiums and claims via damage of household items is usually provided for the mobile banking or airtime. This has reached many entire content of the household. This includes both things millions of emerging customers previously excluded that the owner deems worthy of insurance and those she from conventional insurance markets27. Increasingly, cares less about. IoD offers customers to insure only what this growth (with its success stories and occasional they care about. As additional benefit, it provides users failures) is motivating regulators to discuss frameworks with an inventory of their possessions and their worth. to support responsible development of this market, often with support from the World Bank Group or IoD works because it overcomes one main reason the Access to Insurance Initiative. To achieve their for annual insurance contracts. Traditionally, objectives, these discussions need to include not only establishing an insurance contract caused expenses insurance regulators but also telecommunications that are mostly not proportional to the size (in terms of regulators (because of the crucial role of MNOs) and premium or sum insured) of the contract. The cost of central banks (because of the alternative payment transacting paper based information, transferring it from methods). Stakeholders in these discussions would be paper to IT, assessing the conditions for insurability, well advised to take note of InsureTech developments and invoicing insurance premium carried a minimum outside their jurisdictions which may eventually reach cost that made insurance for shorter durations unviable their citizens, to provide an appropriate framework for or expensive. But new more cost effective payment digital insurance beyond m-insurance, which may turn systems have overcome these limitations, and the use out to be only a transition towards even less traditional of mobile phone based banking makes payment of forms of insurance provision. insurance premium sustainable even on a daily basis in places like Pakistan. Section 3.4 – Insurance on Demand (IoD) Two technologies come together to enable IoD. Insurance on Demand questions why insurance cover More cost efficient payment systems allow to buy has to be purchased in yearly doses. The “sharing insurance for cents at a time. And the omnipresence of economy”, exemplified by AirB&B and carsharing smartphones provides the means to switch insurance businesses, has transformed the concept of property. cover on and off with hitherto unknown ease for both It facilitates the temporary use of things and raises the client and insurer, at marginal administrative cost. question of their need to be permanently insured. IoD Applications on smartphones furthermore allow to add allows people to decide when they want insurance cover new items to the insured list, choose their sum insured to start and to end, and to do that repeatedly during one and deductible, and to post claims and interact with year. For example, sporting equipment can be insured the insurer in their settlement, all at very low cost if only when it will be used during the skiing season. suitably digitized on the insurer’s end. In addition to Telematics – and even simple smartphone apps - allows connectivity, smart phones also bring cameras and to insure cars only when they are driven (also referred to geolocation to the use of insurance. as usage-based insurance), and other ways to insure rarely used cars only when needed are offered by InsureTech This use of technology allows IoD companies like startups. Positioning itself in the context of the sharing Trov to transfer the “sachet principle” to insurance. economy, IoD ventures not only seek to benefit from The sachet principle was Unilever’s answer to the insight a tendency that resonates with new target groups such that the price of shampoo was not the main barrier for as millennials, but also bet on substantial growth of the its use in subsistence marketplaces in India, but the risk sharing economy. Safeshare, “the insurance solution implied by large bulk outlays for unknown products for the sharing economy”, estimates that the sharing combined with liquidity constraints faced e.g. by day economy sector (including accommodation, car sharing, laborers. Shampoo was henceforth sold in increasingly and online staffing) will grow from 15 billion USD in smaller packages, leading to the single-serve sachets 3. INSURETECH OVERVIEW 15 that now constitute 70% of shampoo sales in India.29 self-activation principle. Although distributed through The best known transfer of this principle has been the 2,500 outlets of a retail chain belonging to the same availability of cellphone airtime in small amounts via conglomerate, uptake also fell short of expectations, scratch cards, which unlocked the mobile telephony to and the insurance was eventually discontinued – also even the poorest, and allowed this industry to become because with inflation, the cost of producing the masters of the high volume small margin business physical scratch cards became unsustainable31. model. Insurance has largely failed at replicating this Regulators are discussing potential issues with model so far, in part because of the high administrative IoD. In jurisdictions where rates have to be approved and distribution expenses (the “packaging cost”) of before a product can be marketed, IoD’s possibility any unit of insurance cover30. Because of this, first to “insure anything in a snap”32 will require to re- time buyers of insurance faced two major barriers: examine how rates and conditions can be pre-approved. they often need the liquidity to pay for annual upfront Another worry refers to moral hazard and insurance premium, and they always face a substantial decision fraud which might be encouraged if insurance can which can negatively impact their finances for a whole be switched on and off at will; addressing this with year if based on mis-selling or misunderstanding (and higher premium might lead to an antiselection spiral in attract ridicule by friends and family). which only higher risks or more fraudulent intentions Microinsurance has been familiar with the sachet remain attracted to the insurance. At present there is principle, but has rarely harnessed technology uncertainty about customers’ temptation to switch to take full advantage of it. In 2008, Kenya Orient on insurance after the damage has happened, thus insurance launched the short term personal accident invalidating the entire concept, and this seems to the insurance “Safari Bima” which was purchased on most fundamental challenge for practitioners and scratch cards like those of airtime, and like airtime regulators. Clarity may emerge only as IoD ventures had to be activated by the user (“The Safari Bima scale up and behavioral data accumulates. In any case, scratch cards work like the regular mobile phone IoD makes it more challenging for insurers to keep top up card”). But while it met an insurable need track of their commitments at all times and make sure that had particular demand during the 2007/8 post- they are able to honor their liabilities. election violence, Safari Bima did not reach the While not developed with low income households in anticipated scale, partly because the scratch cards had mind, the value proposition of IoD is highly relevant to be delivered to customers and the economies of scale for inclusive insurance. IoD ventures show how to make were not the same as for airtime. 4,000 policies were insurance purchase decisions “bite-sized”. While the sold, and Safari Bima was eventually discontinued. concept has been around in microinsurance, and low cost In 2007, ACA Insurance in Indonesia offered insurance digital payment systems increasingly make payments of against dengue fever based on the same scratch-card small insurance premium and claims amounts sustainable, A Technology Enabled Example Of The Sachet Principle In Insurance – The Acre Africa Replanting Guarantee Smallholder farmers can achieve much higher yields if they use suitable inputs, and if they can afford them, money spent on seeds or fertilizer generates high returns, unless the crop fails, typically due to climate anomalies. So buying a bag of seeds remains a risky decision for these farmers. Acre Africa combined innovative use of various technologies to mitigate that risk. When farmers buy bags of maize seeds from some agro dealers, they find inside a scratch card similar to those used to top up cellphone airtime. When the farmers plant the seed, they transmit the card’s unique eight digit number via USSD to the scheme operator from their farm. That starts the insurance period, they are enrolled as policyholder with the personal details corresponding to the registration of their SIM card, which also provides their mobile banking connection, and the geographic location of their farm is determined for satellite surveillance of local rainfall. If rainfall is too low for germination during the following 21 days of cover, the satellite data triggers a payment of the corresponding index-based insurance to the farmer’s mobile banking account, allowing her for example to buy another bag of seeds if the rainfall has just been delayed. That way she can catch up with the expected crop production. The entire process is highly digitized and cost efficient. 16 HOW TECHNOLOGY CAN MAKE INSURANCE MORE INCLUSIVE the other technology component – smart phones and the relevant dimensions such as a person’s propensity to corresponding applications – have not yet been widely buy, his brand loyalty or inclination to commit fraud, harnessed for inclusive insurance. With falling prices for requires computing power and algorithms not widely smartphones and their growing availability even among available until recently. Standard statistical tools allowed lower income households, it is not unreasonable to expect UK insurers in the 1990’s to differentiate car insurance that processes tested and matured in IoD will soon be premium based on the driver’s astrological sign34, but transferred to developing country insurance markets, more sophisticated methods were recently used by a where increasing numbers of insurers already offer some UK bank to determine how financial behavior correlates apps to their current customers of conventional insurance with driving behavior and justifies offering discounts to to increase the ease of interaction, customer loyalty, and account holders who stay within overdraft limits. Similar convenience33. Making properly digitized insurance methods recently led a retailer to offer lower insurance available for first time buyers in small “sachets” will not premium to a subset of its customers identified through only overcome liquidity constraints and offer insurance data collected via the retailer’s loyalty scheme35. at overall lower cost; more importantly, it will reduce Correlation is increasingly complementing causality the risk of mis-buying insurance and thereby lower the in our understanding of the world. Most knowledge hurdle to try it, ideally supporting emerging customers we are presented today is of statistical nature, generated with the high degree of client-centric interaction that is by surveys or randomized control trials. It is no longer part of many IoD ventures (albeit often delivered via necessary to understand why someone’s financial chatbots). In addition, IoD is likely to teach insurers and or shopping behavior makes it possible to predict their distribution partners how to bundle insurance with her driving style to use that information in pricing other services more cost-efficiently and transparently, decisions. It is not necessary to assess the impact of opening the door to bundled insurance covers with more weather anomalies on the farm of someone who bought convincing client value. All this may gradually lead index insurance as long as the statistical correlation to insurance becoming part of everyday life, both for between weather index and crop damage is reliable. It millennials in New York and for slum dwellers in Nairobi. is no longer necessary to understand why and how an identifiable condition or behavior relates to insurance Section 3.5 – Machine learning, artificial as long as its impact on insurance claims has been intelligence (AI) and big data established. While car insurance premiums based on Data is said to be the lifeblood of insurance. Data is astrological signs were rejected by the market in the required to understand the target market’s circumstances 1990’s, predictors that might seem equally random are and insurable interests and needs, as well as their used in pricing decisions today (albeit less overtly). willingness and ability to pay. Data is furthermore needed Acceptance of statistically generated knowledge to design and price insurance products that respond is higher than ever before. One downside of this is to these insights, both at initial stage and to monitor that the potential of insurance to motivate more risk- actual versus expected performance of the product so conscientious behavior of customers may erode when as to be able to adjust parameters. Insurers have been the links between premium and behavior become doing this since the advent of computers. What is new increasingly intransparent36. now, however, is the amount of data that consumers in Machine learning has matured and become more developed – and increasingly in developing – countries widely available to extract insights from large produce, and the tools available to analyze this data. amounts of data, at lower cost and much faster than Every interaction with a connected device creates a trail before. AI consists of machines replicating cognitive of information, from habits, preferences, interests, and behavior typically associated with humans. Machine affiliations, to geographic location, purchasing power learning (a subset of artificial intelligence) means giving and spending patterns. Social media, online shopping and computers the ability to learn without being explicitly wearable devices invite ever more people to disclose rich programmed. For years already it has been part of our information about themselves and their lives, resulting in everyday life, in search engines, spam filtering, and the amount of what is called big data. Finding patterns personalized advertising and recommendations37. It is a in these amounts of data and relating them to insurance 3. INSURETECH OVERVIEW TITLE 17 powerful tool to discern patterns from data that may be risks which will therefore move to other insurers; too sparse, noisy, raw, or vast for conventional statistical these other insurers will not easily be able to predict analysis. For example, MasterCard has filed a patent what this shift in disaggregated portfolios implies for application for a method to estimate people’s weight the claims experience, nor how to react to it. That is from their payment transactions and intends to sell that one explanation for the considerable interest of the information to airlines as a basis for passenger seating. insurance industry in artificial intelligence, big data and For insurance, patterns discernable in unstructured data machine learning, but makes it difficult to predict their are of interest when they allow to predict the likelihood mid-term impact on insurance. or severity of claims and the likelihood of fraud. In Advanced understanding of risk relevant indicators, theory, they could also predict price elasticity of an along with underwriting processes supported individual’s demand for insurance or his inclination to by artificial intelligence, allow to streamline the change insurer if service is not prompt. Such predictors insurance enrolment process and make it more allow a higher pricing granularity, known for example convenient for the customer. A better understanding from different life insurance rates for smokers and non- of risk factors relevant to a specific person and smokers. But they carry the risk of eroding solidarity, product can reduce the standard number of questions and while some predictors of higher insurance cost can traditionally asked to assess insurability of a risk. If the be addressed by the individual – stop riding motorcycles insurer already has access to additional information on – others such as inherited conditions cannot. the applicant, less questions may need to be asked. If Advanced understanding of risk relevant indicators automated underwriting systems are employed which will lead to more granular pricing and put pressure on can interactively prompt for additional information insurers who don’t follow. Advanced analytics on ever or specifications (see 4.), the time it takes to buy an larger amounts of data allows insurers ever more precise insurance can be reduced from weeks to minutes. This predictions of an individual’s expected insurance claims is the proposition e.g. of HavenLife, a venture backed amounts. This allows for differentiation of premium by MassMutual which justifies its slogan “getting accordingly, and for offering insurance to some that life insurance should be painless” with simple online is not made available to others (within the regulatory applications and immediate decisions. AI is also behind boundaries). Furthermore, when embedded in suitably the chatbots that provide “robo-advice” under the digitized environments, such decisions can be taken Concierge Distribution model. instantaneously by software. But every disaggregation AI can streamline insurance processes in new ways. of a risk pool – for example, disaggregating all men in One area where AI is having increasingly noticeable those who smoke and those who don’t – leads to more impact is natural language processing and speech-to- favorable conditions to one subset and less favorable text conversion. Computers can sustain dialogues with conditions to the other. Smokers will therefore switch customers that resemble the traditional insurance sales to insurers who don’t differentiate premium by nicotine process, but can do that over the phone, in a multitude use. These insurers will see their claims increase with of languages, at large scale and low cost, while the increasing proportion of smokers in their portfolio. abolishing the need to fill forms. Insurify, a Concierge If they respond by increasing premium rates, non- Distribution venture, includes a virtual advisor that can smokers will leave and the proportion of smokers will talk to people in natural language. Not only does this further increase. Hence, they have little choice but reduce cost and increase turnaround time, it also has to differentiate premium by smoking status as well. the potential to improve customer experience as these While that was easy for smoking / non-smoking as the approaches mature. In most cases, AI applications are cotinine test was robust and affordable and the excess expected to reduce operating cost, as tasks traditionally mortality was well researched, machine learning on big requiring highly specialized humans are delegated data is producing proprietary intellectual capital that to algorithms. Media coverage of resulting staff other insures cannot easily replicate. In other words, redundancies at insurers started in January 201738. one insurer’s proprietary AI can uncover a risk-relevant differentiator that helps it be less attractive for higher 18 HOW TECHNOLOGY CAN MAKE INSURANCE MORE INCLUSIVE But artificial intelligence can also allow insurability Discrimination can be another concern. Unless the owner of of new risks in new circumstances. One example the AI algorithms who determine who gets what insurance is the International Food Policy Research Institute’s at what conditions disclose these algorithms to insurance “Picture Based Insurance”. It is testing an alternative supervisors (and insurance supervisors are capable of crop loss assessment that can make agriculture understanding them), the risk of racial profiling or other insurance sustainable for small scale farmers. It uses forms of discrimination proscribed by insurance regulation the smartphone cameras and a special application to and society is difficult to manage. Such disclosure is collect visual information on crop development by questionable at present, considering the substantial local farmers, without the need for experts to visit the investment in this intellectual property and the competitive (often remote) fields. The application also provides an advantage it promises. On the other hand, dynamic and efficient way to collect additional information from tailor-made rating on applicants may not be possible where the farmers, and a combination of machine learning jurisdictions insist on traditional actuarial justification of and expert panels aims to improve the reliability of insurance premiums and conditions. these data to predict crop losses. Analytics based on Another issue is the legal uncertainty surrounding artificial intelligence applied to big data could lead to professional advice provided not by humans (such as considerable progress in the insurability of risks related licensed and trained insurance intermediaries) but by to pandemics, disasters, disruption of global supply machines. In January 2017, the European Parliament chains or climate change. The insurer AIG is combining started discussing a legal framework for robotics and AI AI with wearable devices to reduce the risk of workplace , but it will take considerable time for jurisdictions in injury. Analyzing vast amounts of unstructured data developed and developing countries to regulate AI. from multiple sources also allows insurers to better And while these efforts are underway, AI will continue detect changes in society and environment that can to evolve, making up-to-date regulation increasingly impact insurance utilization, thus reducing the risk of difficult. In the meantime, supervisors have to decide change that made some risks difficult to insure. AI can on the liability aspects of robo-advice and how much facilitate claims assessment by judging the likelihood automation of insurance processes they accept. of fraud from other data, reducing claims adjustment Suggestions to point out more clearly when customers are costs that might have been prohibitive before, and it communicating with a machine and not another person can add precision for claims reserving. may obliterate some expected advantages especially Regulators are faced with a number of questions in customer experience, if it makes customers uneasy. concerning AI and its uses in machine learning and In addition, ever more granular underwriting, and products analyzing big data. Data protection is a considerable tailor-made to ever smaller target sub-markets, may erode concern: mining data from social media, online purchase the tenet that insurance is transfer of money from the records, financial and other records to assess an individual’s fortunate to the unfortunate, questioning the very solidarity risk conflicts with privacy rules in many jurisdictions, and principle that has been core to insurance. This problematic is often not acceptable to the general public. Furthermore, trend could go largely unnoticed where the awareness of the the data collected and generated through analysis will likely solidarity principle is low, especially in developing countries reside on cloud servers beyond the control of the insurer (in with low insurance literacy. This trend can also lead to an case he partners with an AI company) and the insurance atomization of risk pools that threatens the applicability of supervisor. Although surveys with, there are worries that the law of large numbers, another pillar of insurance that is they might regret this disclosure at a later date when it leads often misunderstood. not to lower but to higher insurance premium or outright And to the extent that AI tools not owned by insurers refusal. Some regulators think such disclosure decisions become increasingly determinant to competition, require similar protection as other aspects of insurance insurers may find themselves relegated to providers of purchase decisions. The constant surveillance envisioned license and balance sheet, losing the client relationship from harnessing the stream of risk-relevant information to technology companies not supervised by insurance from wearable devices (as well as telematics and the Internet regulators, and vulnerably dependent on the partnership of Things) will require broad societal debate that may result with the AI provider. in its rejection (and possible clandestine continuation). 3. INSURETECH OVERVIEW 19 What is clear is that more widespread use of AI in and can reduce mis-selling through standardized (if insurance confronts regulators and supervisors complex) and auditable decision trees. AI can facilitate with new challenges, and may drive them to adopt the identification and authentication of people based more technology-based approaches to regulation on voice or facial recognition, improving access to themselves. This might be necessary for example to insurance to people without formal IDs; interpreting evaluate the data used by insurers for product design, facial expressions can also help against insurance pricing and reserving decisions, and to assess the fraud. AI furthermore allows to detect insurable needs algorithms used in this and in other areas of insurance from patterns in big data, and to generate calculatory operations such as investment, asset-liability bases to price and reserve products meeting these management and risk management. With the advent needs even where traditional approaches to insurance of FinTech, supervisors of other financial sectors would have struggled with the absence of statistics – a have been faced with this challenge before, leading common worry in microinsurance. Both innovative and to coin the term “RegTech” to technology-supported conventional products can be tailored very specifically supervisory answers to technology-driven financial to individuals or businesses, so that they provide better innovations. While several companies already offer value for money. Pricing and product design could be products in that space40 and jurisdictions are starting adjusted quickly to respond to new insights, lowering to compete in this respect, a market for insurance- the hurdle to launch new products in previously specific solutions has not yet emerged. data-scarce situations. Completely new distribution channels can be harnessed when all insurance expertise The possible applications of AI in inclusive insurance necessary for the sales and enrolment process resides are manifold. AI applications are expected to reduce in a software that easily interfaces with other systems. cost of insurance significantly, making it more affordable Some of these channels will be the owners of big data to emerging customers. Despite this cost reduction (that generated by their clients e.g. from online sales, and includes intermediation cost like agents’ commission), will thus re-define the interest of the various players thorough interactive needs assessment and advice in owning and controlling the insurance. That can can be provided to customers, already capturing the challenge the traditional role of the insurer, but will data traditionally submitted on paper forms. This will also engage distribution channels more forcefully than increase convenience (e.g. turnaround time in every in past attempts to sell microinsurance via microfinance interaction) for everyone who doesn’t mind to discuss institutions (for example). insurance needs with an algorithm (or doesn’t notice), Joint Ventures Between Technology Companies And Insurers – The China Example In January 2017, the Chinese e-commerce company TenCent announced that it was taking stakes in the Hong Kong operations of British insurer Aviva, together with the hedge fund Hillhouse Capital (which also holds stakes in AirB&B and Uber). TenCent, one of world’s top 10 most valuable companies, is the world’s largest gaming company and one of the largest Internet companies, and its services include e-commerce, social networks, instant messaging, web portals, and WeChat which does all of the above for 767 million active users per month. The intention of the deal is to increase Aviva’s digital focus and grow TenCent’s presence in the insurance business beyond its participation in the insurers Zhong An and He Tai. Hillhouse Capital was also part of the joint venture between the German insurer Allianz and the Chinese internet company Baidu in 2015 to establish a nationwide digital insurer after talks with AliBaba, TenCent’s longtime rival for internet dominance, did not flourish. According to Allianz’ press release, “Baidu, a technology-based media company, is the leading Chinese language Internet search provider with a mobile search user base covering over 640 million monthly active users”, and the deal aims to put Allianz at the forefront of disruptive innovation. AliBaba has meanwhile established a partnership with French insurer AXA. Although AXA has considerable market presence in China already, it expects that the partnership with AliBaba and its online payments company Ant Financial will speed up its outreach to more customers. For AliBaba, on the other hand, the partnership is part of the company’s globalization strategy. 20 HOW TECHNOLOGY CAN MAKE INSURANCE MORE INCLUSIVE Section 3.6 – Wholehearted digitization, and channels and even social media, but expect high quality the “digitally born insurer” customer-centric websites and applications, seamless interaction across media, fast response, and high degrees What if all the latest technology would be applied by of data and privacy protection. As more and more an insurance undertaking not encumbered by legacy industries strive to live up to these expectations also in systems and procedures? The outcome of this thought developing countries, even first time buyers of insurance experiment is sometimes referred to as the digitally will be less satisfied with insurers offering pre-digital born insurer, a metaphor for utmost digitization. It customer experience only. And in addition to evolving refers to the conventional business model of insurance consumer demand, evolving regulatory demand also puts – not necessarily P2P or blockchain-based – and is stress on outdated administration systems in insurance. relevant especially when comparing with the insurance industry’s current implementation of technology. The Adopting updated technology could result in substantial majority of large insurers are decades old or older. The cost reduction for the average insurer. BCG and Morgan industry invested substantially in IT in the 70s and 80s, Stanley estimate that combined ratios could be reduced by when most IT solutions for insurers were tailor-made. 17 to 21 %; in detail, the net impact of different technologies Many insurers run a diversity of administrative IT for a motor insurer in a mature market is estimated as systems in parallel, frequently with patches to address • Using Big Data analytics to improve pricing: 6.7 specific aspects rather than state-of-the-art systems percentage points reduction in combined ratio infrastructure, explaining costly and slow processes. • Automating enrolment procedures: 0.1 ppts Relative to revenue, insurance invests half as much • Shift sales to online digital channels: 6.0 ppts in IT as banking41. There is considerable pent-up • Automating service and administration procedures: 1.1 ppts investment need, but the older and larger an insurer, the • Improved fraud detection with Big Data and AI: 1.5 ppts more difficult a radical modernization is. • Automated claims management: 1.2 ppts Consumer demands have evolved beyond the level • Lower reserve volatility using Big Data and AI: 0.1 ppts of service the average insurer’s IT can provide. • Optimizing support functions: 1.2 ppts Consumers in mature insurance markets have grown • Replacing legacy systems with up-to-date front and accustomed to interact with providers through digital back end systems: 0.7 ppts Zhong An, The Example Of A Digitally Born Insurance From China Zhong An was launched in 2013 as China’s first online-only insurer, and posted 25 million USD net profit two years later. In terms of insurance penetration, China in 2014 ranked 57th (between Saint Vincent and India) with total premium amounting to 3.16% of GDP. But between 2014 and 2016, Zhong An sold 5.8 billion policies to 460 million customers. Over 300 million policies were sold on November 11, 2015 alone (the Chinese shopping holiday known as Singles Day). It helps that Zhong An is backed both by insurance conglomerate Ping An and by e-commerce titans AliBaba and TenCent, who enable the sale of large numbers of small policies – for example covering the shipping cost of returns at an average premium rate of 3% – seamlessly and conveniently during online shopping. Starting unencumbered by legacy systems in a regulatory environment that encourages innovation, Zhong An runs all processes and Big Data analytics on a cloud based platform, and an open platform to facilitate seamless integration with the systems of its 300+ partners. This allows scenario based product design to follow client needs identified by partners, allows to tailor products profitably also for small market niches and to sustainably sell policies with very small premium, and being quick to market and to adjust as needed. Zhong An is able to price more accurately, assess risk with greater precision, and underwrite dynamically in microseconds. It has perfected the integration into a multitude of systems across the travel, e-commerce, motor, health and other sectors, and systematically facilitates cross selling. And Zhong An claims to have optimized fraud detection while providing fast, transparent and convenient claims service through substantial automation. All this allowed the company to test its model with somewhat conventional insurance such as shipping return cost and flight cancellations. That was the starting point for the launch of more than 200 insurance products, some quite novel such as health insurance that incorporates information from wearables, drone insurance, or insurance that pays when temperatures exceed 37C. Zhong An, which has been described in the Financial Times as “a technology company that happens to focus on insurance”, exemplifies a business model that is not product centric but customer centric, and built on online ecosystems to respond to customer demand and enhance customer experience. 3. INSURETECH OVERVIEW 21 While this assessment is hypothetical, the report42 insurance intermediated by mobile network operators also analyses the case of two US insurers (Geico and (see paragraph on m-insurance on pages 14 and 15). As Progressive) who have achieved considerably lower an insurer employs one or more of the InsureTech themes expenses through strategic use of IT, funding increases in discussed above like Zhong An, the corresponding marketing budgets that resulted in above-market growth. regulatory concerns highlighted in Sections 3.1 to 3.5 also apply to a digitally born insurer or one transitioning Insurers don’t need to develop all this capacity in- towards that model. Insurance regulators may be house but can buy services that move them towards concerned, for example, that substantial use of AI on Big new business models that support more inclusive Data may lead to pricing and underwriting decisions that insurance. Companies like Shift Technology and they fail to understand and that may discriminate certain TycheRisk offer AI services for claims management consumer groups in undesired ways. to insurers. Companies like SimpleSurance offer seamless integration of cross selling at e-commerce “Digitally-born” insurance, or insurance provision checkout. Companies like CarpeData offer insurers inspired by the example of Zhong An, can contribute predictive scores to “assess risk at critical steps in the in a number of ways to overcome barriers to inclusive insurance policy lifecycle”, and connected platforms to insurance. Lowering cost of quality insurance by “consolidate and functionalize the next generation of lowering administration expenses through determined data”, including data aggregated from social networks. automation of processes and reducing fraud through Such services support step-change towards consumer AI will make insurance more accessible to low income centricity. However, to reach the full transformational households and MSMEs. Cloud based platforms potential of new technologies, incumbent insurers have built around application program interfaces support to question every aspect of their business model. This cost- and service-efficient integration of a variety of can alert “corporate antibodies” opposed to change, alternative distribution partners that may reach people which is why some insurance groups nurture promising not served by the traditional microinsurance multipliers technology investments in separate incubators. like microfinance institutions and cooperatives. These two measures will make provision of insurance with Regulatory frameworks for e-commerce and digital very small premium sustainable even at lower scale provision of insurance are the most relevant for the than that normally required for microinsurance. outlook of a digitally-born insurers, or one moving To the extent that an insurer applies the InsureTech towards “de-materialization”. In some jurisdictions themes discussed in 3.1 to 3.5, the corresponding benefits the frameworks already support this approach, while in respect of inclusive insurance proposed there apply. others are in the process of drafting and discussing But in addition, what sets a truly (or aspiring) digitally regulation, often motivated by the recent growth of Protecting Nascent Innovation From Corporate Antibodies – The Lumenlabs Example MetLife, the largest US insurer (Forbes 2017), launched LumenLabs in 2015 as its disruptive innovation center in Singapore, far from its corporate headquarters but close to vast markets of populations leapfrogging older technologies to go directly digital. LumenLabs “applies a thoughtful, structured and methodical process to incubating ideas, and developing new businesses that can scale”. The process, which distinguishes the exploration phase (frame, explore, create) from the experimentation phase (blueprint, experiment, pilot), is applied to pursue increases in revenue (new revenue through new services, new business models through disruption), cost reduction (improved workflow efficiency, improved claims handling and fraud detection), increased value of new business (new solutions for underserved Asian populations, technological advantage for MetLife intermediaries), and increased persistency (technology facilitated cross selling, sticky value added services). Once an issue has been defined, a call for proposals invites InsureTech companies from around the world to suggest solutions. In one case, 40 submissions were longlisted out of 140 (including one from Afghanistan), and 2 were finally selected. The collaboration with the AI company Shift Technology which resulted from this process saved MetLife Japan considerable amounts of fraudulent claims payments in 2016. More importantly, LumenLabs is subtly transforming the way MetLife staff cooperate with InsureTech startups. 22 HOW TECHNOLOGY CAN MAKE INSURANCE MORE INCLUSIVE born insurer apart is that innovations are integrated in The following table summarizes the potential that the an appropriate architecture and business model, instead various InsureTech themes discussed in Section 3 offer to of being deployed alongside conventional processes overcome barriers to more inclusive insurance discussed and customer relationship paradigms. in Section 2. The corresponding risks will be discussed in Barrier to inclusive InsureTech solution insurance Limited • P2P promises to reduce cost of insurance through disintermediation and less fraudulent claims Purchasing • Blockchain-based models may reduce cost through disintermediation, and smart contracts Power can reduce administrative cost even under conventional business models • Concierge Distribution reduces cost through automated intermediation and advice, and through more transparency that leads to more competition and prevalence of best-in-class providers • Insurance on Demand makes payment of premium easier • Wholehearted digitization reduces operating cost • Artificial Intelligence and Big Data analytics reduce cost of fraud Limited • IoD facilitates trying insurance to confirm understanding Understanding • P2P concepts help first time buyers understand formal insurance in the terms of the informal risk management mechanisms they are familiar with • Concierge Distribution reduces the cost of communicating with clients to reiterate messages, clarify doubts and engage with clients Limited Trust • P2P aims to bring trust back to insurance (including through greater transparency) • Blockchain aims to replace the need for trust with tamper-proof transparency and auditability • IoD facilitates trying insurance to test reliability of providers and products • Concierge Distribution positions itself as provider-independent partner and advocate of client • Wholehearted digitization and AI allow faster response including payment of legitimate claims, improving tangibility Unsuitable • P2P models can crowdsource product needs proposals from their members; Concierge Products Distribution also offers that potential • AI and Big Data allow to identify previously hidden insurable interests • AI and Big Data allow to generate the calculatory bases to design and price new products even when traditional statistics are missing • AI and Big Data support underwriting and claims handling of products and customer groups that traditionally could not be underwritten, or only at high cost • Consumer experience – an integral element of insurance products – is enhanced through appropriate processes supported by elements of Concierge Distribution, IoD, or wholehearted digitization • Wholehearted digitization can reduce operating cost to a degree that previously unsustainable products become viable Unsuitable • Smartphone based, disintermediated distribution via Concierge Distribution and IoD Distribution can sustainably reach new segments where agents and brokers (as well as the typical microinsurance distributers) failed to reach scale • P2P relies on social media word of mouth (potentially “viral”) dissemination • With cloud based platforms and application program interfaces, wholehearted digitization facilitates seamless integration with a new generation of potential distribution partners, for example in the mobile and e-commerce sectors • Wholehearted digitization and dematerialization / paper-free insurance overcomes geographic barriers (paper trail logistics) Unsuitable • IoD has the potential to make the large-volume-small-margin business model work for Business insurance, making provision of small insurance sustainable Models • Wholehearted digitization, and digitally-born insurance in particular, reinvent the business model towards customer centricity and cost efficiency • Increasing internet use and smartphone ownership among low income households bring them into a potentially global market for digital-based insurance where their socioeconomic status is less relevant 3. INSURETECH OVERVIEW 23 24 HOW TECHNOLOGY CAN MAKE INSURANCE MORE INCLUSIVE 4. Risks Radical change is part of the unique selling proposition of many InsureTech startups; if just some of them succeed on a scale commensurate with the funding they attract, insurance and the role of incumbents will change in ways that are difficult to predict. Despite occasional new products responding to new societal circumstances and the occasional adoption of omnipresent technologies, provision of insurance has changed little since the 19th century. Although donor funding has allowed for testing of a variety of new approaches in microinsurance, they have not contributed much to disruptive innovation in conventional insurance, even in the areas and countries where they scaled up successfully. It is this pent up innovation potential in insurance compared to other sectors that is attracting investors. One immediate risk is posed to incumbent insurers. Depending on the InsureTech model, their role may be relegated to providing the license and balance sheet while client ownership and interaction will be held by other players. Some insurers may be replaced by new entities like digitally-born insurers or P2P and blockchain structures (albeit some possibly operating in regulatory grey areas for some time). While in jurisdictions with strong legal frameworks and supervisory capacity, “regulation is the one problem you cannot solve by throwing money at it”43, jurisdictions with weaker supervisory capacity may take a while to formulate a response to some proposals like pure P2P ventures. This can lead to an uneven playing field where incumbent insurers feel disadvantaged and have an even stronger motivation to wish failure to InsureTech ventures. Regulatory “sandboxes”, i.e. clearly defined frameworks where startups and incumbents alike are given protection from some elements of regulatory uncertainty for a limited time and under close surveillance, are one tool to address this situation which is being deployed not only in Europe (e.g. UK) but also in Asia (e.g. Singapore, Malaysia), sometimes supported with innovation hubs that facilitate early stage dialogue with regulators. But even with level playing fields, market structures are likely to change (to the benefit of the consumer, according to the promise of InsureTech companies). One likely change from increased automation of insurance processes is a reduction of the number of people earning an income from work in insurance. Even where InsureTech startups emphasize that they are fundamentally different from conventional insurance, they may tarnish the reputation of the insurance industry if they fail spectacularly. Statements like Ariely’s “If you try 4. RISKS 25 to create a system to bring out the worst in people, can learn from the evolution of m-insurance and you’d end up with one that looks a lot like the current consider adoption of concepts like the living will insurance industry” and claims that “old insurance that assures orderly unwinding of a partnership or is rubbish”44 disparage the image of conventional enterprise46. Another lesson from m-insurance is insurance, and the justification for many InsureTech that structures that rely on a functioning partnership startups - the perceived dissatisfaction of consumers between various parties are vulnerable to one of the - is amplified in their marketing. Whether or not such parties abruptly leaving the partnership. This risk criticism is accurate, confidence in insurance may end will likely be higher in InsureTech, because more up being lower unless InsureTech can live up to its core functionality is outsourced beyond the control promise to replace old insurance with something better. of the insurer to enterprises with a unique selling Confidence can be particularly shaken if an InsureTech proposition that cannot be easily replaced and who venture was to fail very visibly, especially if it had contribute proprietary technology that cannot be easily attracted many customers already. Failure in this case reproduced by the insurer if necessary, augmenting the does not necessarily mean insolvency and default question of ownership of key business assets. on obligations towards clients, but could be merely More systemic risks, however, are not likely to a withdrawal from the market due to disappointed originate from InsureTech in the near future, (investor) expectations. With the number of startups, especially from inclusive InsureTech, due to its this is a realistic scenario: Munich Re, for example, currently limited scale. The most urgent insurance sees 50-100 investment proposals each month, and needs of people currently uninsured (in developing know that some of their investments in this space countries) or underinsured (in developed countries) will fail. Munich Re has a plan to manage failures so can be met with short term insurance without elements that no customer gets disadvantaged and regulatory of capital accumulation. By contrast, the biggest risks requirements are met45, but that may not be the case for to the insurance sector – which have the potential every investor everywhere. to threaten an entire economy’s financial stability Reliance on innovative technology that has not – so far related to long term commitments that were been vastly tested poses a threat to the operational not correctly anticipated, such as asbestos liability stability of insurers in general and InsureTech claims or interest rate guarantees. Increasingly, ventures in particular; even when their disruption capital requirements reflect these risks, motivating or failure does not tarnish the reputation of the incumbent insurers and startups alike to avoid them. insurance industry, it can harm customers. Some This potential long term nature of insurance liabilities, InsureTech ventures are based on latest technology and the resulting risk of asset-liability mismatches, are such as blockchain or artificial intelligence that has not not palatable to most investors in current InsureTech been widely tested so far. Nor has their integration with ventures, who will therefore steer clear of them. While the traditional procedures of insurance provision been long term financial planning and asset accumulation tested, or the novel business models based on them. is a growing need in developing countries and among All ventures assume reliable connectivity, interface low income households, the financial planning horizon standards, electricity supply, data security and the there does not yet lead to noticeable demand for availability of specialized human resources; these long term life or annuity insurance. Furthermore, the may be reasonable assumptions in very developed predominant investment strategies there respond to countries but less so in developing countries. On the emerging local capital markets, and risks e.g. of the other hand, insurance markets there are less concentration and related party transactions outweigh constrained by tradition and are cognizant of greater possible risks of InsureTech related market changes need for innovation. Beyond reputational damage, the such as disruption of new business of incumbents. risks to consumers may be lower than in FinTech, as On the other hand, InsureTech elements contribute insurance for emerging customers will continue to be to mitigating some of the traditional risks of short term and without asset accumulation in the near insurance undertakings. The general expectation to mid future. But abrupt loss of health or life insurance that InsureTech will allow more accurate pricing cover can still lead to hardship. Insurance supervisors 26 HOW TECHNOLOGY CAN MAKE INSURANCE MORE INCLUSIVE and underwriting, through Big Data and Artificial some maturity and the risk of technical malfunctions Intelligence for example, should lead to a lower is outweighed by the benefits of conscientious underwriting and pricing risk. Big Data and Artificial reengineering of processes. Intelligence can also help reduce the product design A detailed assessment of the risks specific to a risk, as well as the risks of incorrectly forecasting the particular market is advisable when InsureTech economic environment and policyholder behavior. ventures or themes start spreading there. Insurers Claims risk can be reduced through improved will have to revisit their enterprise risk management, detection of insurance fraud. Increased automation and regulators together with other stakeholders will and digitization, as well as more immediate customer have to evaluate a more detailed and country-specific engagement, allow for prototyping and testing of new list of potential risks, including regulatory arbitrage, products more efficiently, and help companies to react mis-selling, money laundering, data and privacy quickly when deviations from expectations suggest protection, market dominance and the crowding out changes to products or processes. If done properly, of established insurance industries, systemic threats to increased automation and digitization are also likely to financial market stability, and discrimination. decrease operational risk, at least when models reach Data Protection And Data Privacy New ways to process old and new types of (meta)data are at the core of most InsureTech propositions, and given the nature of insurance, most of the data in question is personal. Rules concerning ownership, privacy and appropriate handling of customers’ personal data exist in most countries. But they have evolved in times where the majority of personal client data was disclosed by the client herself via paper based processes (such as insurance application forms) that put limits on the amount of data to be captured. Labor intensive procedures limited the use of this data. Today’s data privacy and protection rules don’t fully reflect the increased amounts of data generated by the use of mobile phones, social networks, media streaming and online shopping, and they don’t reflect the new possibilities to make sense of these increasing amounts of data, for example by combining anonymized customer transaction data from several sources to guess the real identity of the corresponding person and henceforth target him with customized offers. With cross selling being a frequent objective of InsureTech ventures, exchange of customer data between different businesses and via the InsureTech third party will increase. Open application programming interfaces, an enabler of many InsureTech ventures, open new gateways for the proliferation of personal data. Social networks and other dominating technology companies have made growing numbers of people increasingly comfortable with the sacrifice of privacy required for full convenience of online experiences. Market research shows that the “oversharing generation” of millennials is particularly willing to provide details about their personal preferences and habits to marketers, in exchange of the smallest rewards. In the case of insurance, such rewards can come in the form of lower premium, and can be substantial in the short term. But it is unclear what conclusions will be drawn by algorithms from peoples’ data as time passes, and initially lower premiums can turn to higher premiums (be it reflecting perceived higher risks or lower price sensitivity), or exclusions. Being widely acceptable to one socioeconomic group, user agreements and terms of service that require customers to relinquish some data privacy may increasingly be imposed on other customer groups as well. Another issue regarding personal data is its security. Large scale data breaches involving retailers like Target and insurers like Anthem highlight the vulnerability of data to criminal attacks, and raise the question of how well small InsureTech startups are able to protect the data they handle. On the other hand, many jurisdictions require that personal data – for example related to insurance – is stored on servers within the geographic boundaries of the country. That presents a considerable obstacle to cloud-based IT solutions, challenging the viability of some insurance innovations. Regulators across different competences (like e-commerce, telecommunication, and finance) will need to find an appropriate balance between fostering innovation and protecting consumers. 4. RISKS 27 28 HOW TECHNOLOGY CAN MAKE INSURANCE MORE INCLUSIVE 5. Recommendations A better understanding of InsureTech will help players along the insurance value chain in developing markets relate the trend to their circumstances. InsureTech being a relatively new phenomenon, the familiarity of insurers and their potential partners in developing countries with it is still limited. Investments in InsureTech were geographically more diversified in 2016 than in previous years, but still 62% went to North America, and another 16% to European countries. Incubators, innovation hubs, investor meetings and conventions are only just starting to spread awareness and understanding of InsureTech in Africa and most of Asia and Latin America, even though exciting innovations are being piloted there which perfectly fit the label47. So while some insurance innovation happens everywhere, the movement that attracts attention of investors, entrepreneurs, regulators and the media, and spreads understanding, is yet to arrive to developing markets. Helping everyone understand how emerging solutions can help serve emerging customers in emerging countries will accelerate the transfer. Local practitioners along the insurance value chain may appreciate help to understand what InsureTech will bring to their markets. Even when InsureTech concepts are well understood, it is not straightforward to predict how they will affect a developing insurance market and the business of any one insurer, intermediary, or specialist professional such as underwriters and claims adjusters. Insurers and brokers that are part of an international group with strong presence in North America or Europe may get guidance and even strategic direction internally, and large international insurance and reinsurance groups mostly have their strategic approaches to InsureTech (such as MetLife’s LumenLabs). But purely local players may lack insights to anticipate how InsureTech could affect competition and client behavior, and how they can prepare and participate. Regulators will be faced with new questions that will demand fast response, in coordination with the insurance industry and regulators of other areas. M-insurance provides an example where regulators in Africa and elsewhere have welcomed support – for example in form of organized South-South exchanges – in adapting the frameworks to new realities. The need for such support in maintaining a fair, safe and stable insurance market will be even greater when more innovative forms of InsureTech arrive. 5. RECOMMENDATIONS 29 The same practitioners may benefit from 300 million persons, and index based agriculture encouragement and guidance to address barriers insurance. Admittedly, not all donor-funded ventures to inclusive insurance with elements of InsureTech. in microinsurance proved successful, but the lessons Even though much of InsureTech is motivated by they generated have accelerated the progress along the voluntary underinsurance by certain groups in developed learning curve of others who today look successful. countries, it does not generally focus on low income Donor funding can thus be particularly impactful individuals or MSMEs. InsureTech does offer clues on when it comes with the commitment to document and how to overcome exclusion and serve more low income disseminate lessons learnt. This will be no different households and firms. But expertise is required to make in the effort to harness InsureTech for more inclusive that transfer, especially until successful InsureTech insurance in markets where exclusion sustains poverty. companies come to developing countries, and offer The World Bank (WB) has a unique role to play, their services there guided by global and local lessons for example in respect of policy and regulation. It learnt in microinsurance. Until that time, concepts such has a long track record of helping governments shape as Concierge Distribution have to be recreated by local conducive policy for finance and other areas, and can companies, along with the underlying technologies provide financing to implement necessary modernization. and business processes. That should be easier than Examples include the Financial Inclusion Strategy recreating AI for example, but would still require going Framework program and other initiatives for financial through the learning curve with little more guidance inclusion, consumer protection, and financial literacy. than lessons made public from pioneers elsewhere. It has equal expertise in helping jurisdictions assess Given the so far moderate interest of most insurers in and improve their regulatory frameworks for insurance developing countries to venture beyond the comfort in accordance with government policy, particularly in zone in order to better serve the poor, InsureTech will not respect of new developments such as microinsurance lead to innovative inclusive approaches without some and index insurance. Long standing close dialogue handholding. In this respect, the insurance value chain and cooperation with standard setting bodies such as also includes regulators and supervisors; they will need the International Association of Insurance Supervisors to balance the usual consumer protection considerations or the European Insurance and Occupational Pensions with encouragement of inclusivity through innovation, Authority ensure that the WB is informed of latest and they will need to learn how to effectively supervise consensus of good practices, and in fact contributes to operations based on new technologies. shape them. That has allowed the WB to help regulators Donors can support the transfer of InsureTech in many jurisdictions around the world to update their concepts to inclusive insurance. InsureTech is regulatory frameworks (in dialogue with the industry) so attracting considerable amounts of investor funding that they better reflect recent developments in insurance, – 1.7 billion USD across 173 deals in 2016 – but including microinsurance, index insurance and mobile that goes mostly to developed insurance markets. insurance. It also makes the WB a trusted partner when Some multinational insurance groups proactively updating supervisory capacity to new circumstances. The explore how InsureTech can improve their outreach in WB also contributes to disseminate the understanding developing countries. But most local insurers and their of new rules among stakeholders, partnering for partners will be more ready to explore innovations example with the Access to Insurance Initiative. for inclusion with external support. That support can As InsureTech spreads, the WB can help jurisdictions be in the form of technical assistance – bringing for find appropriate approaches to licensing and supervision example specific skills and lessons learnt elsewhere – requirements of new types of players along the insurance and in the form of suitable financial investments, for value chain that is getting longer. And it can help to example in startup joint ventures. It can also come in harness technological innovation to make social security the form of public goods, for example assistance to systems more inclusive. Technology, for example, to regulators to implement a suitable framework, possibly make adherence of informal sectors to national universal including “sandbox” or similar approaches. Substantial health care schemes easier, and increase the efficient donor support has contributed to the development operation of these systems for improved client value. of microinsurance, which today covers more than 30 HOW TECHNOLOGY CAN MAKE INSURANCE MORE INCLUSIVE The WB’s Finance, Competitiveness & Innovation markets and in new technologies which support the Global Practice also contributes to strengthen the insurance industry such as payment systems, mobile wider ecosystem required for inclusive insurance to wallets, and new business models like aggregators and flourish, with interventions that go beyond policy digital insurers. It supports its investees in the adoption and regulatory advice. A good example is technical of new technologies and approaches to reduce poverty assistance to leverage agent networks established and increase prosperity, and partners with champions for mobile money and other digital financial of impactful technology such as Ant Financial to enable services for distribution of microinsurance products. digital financial inclusion in Emerging Markets. The Another example is support to bundle insurance IFC has investment and advisory services expertise with governmental cash transfer programs for better in a wide range of areas that include venture capital, achievement of policy purposes48. Beyond payment FinTech, digital financial services, microfinance systems, conducive infrastructure also includes national and microinsurance, and health. The IFC’s Gender identification systems, and sustainable availability of Secretariat has particular expertise in making insurance data that can be used for insurance and other purposes, work for women – both as service and as employer and the WB can help countries identify gaps and – which it provides to its partners. The Global Index develop action plans to close them. Such technical Insurance Facility has delivered proof of concept for assistance, provided by specialists experienced in a various business models for agriculture and disaster variety of fields, can be supported by donor funded index insurance that have reached millions and trust funds, or it can be part of larger interventions that provided guidance for many other ventures, piloting deepen and broaden countries’ financial sectors, which technologies such as satellite-based payment triggers. may include components for specific implementations. The IFC therefore plays an important role in the Donors who want to have an impact in any area of development and dissemination of new ideas, markets financial services can consult with the WB to identify and services. the most promising strategies. One example of the The Consultative Group to Assist the Poor (CGAP) WBG’s many partnerships is the collaboration with the contributes to this work, promoting financial UK Department for International Development under innovation and customer centricity in particular. the Harnessing Innovation for Financial Inclusion Housed at the WB, CGAP is a global partnership of program that focusses on the use of technological over 30 leading organizations that seek to advance innovations to deliver financial services to some of the financial inclusion. Its mission “to improve the lives world’s poorest and most excluded people, particularly of poor people by spurring innovations and advancing women and people living in fragile and conflict-affected knowledge and solutions that promote responsible, areas. This multi-dimensional work with standard sustainable, inclusive financial markets” is particularly setters and practitioners explains the convening power aligned with the drive to employ InsureTech for of the WBG, which is welcome by the G7 and G20, greater financial inclusion. CGAP’s research provides the UN and the Alliance for Financial Inclusion, to insights into the financial lives and needs of low support both high level dialogue and the advancement income populations, and how well they are served by of tools, standards and solutions. The WB uses its traditional and new financial services such a mobile convening power also to foster South-South dialogue, insurance. These insights are used to help policymakers communities of practice and other forms of knowledge and regulators better calibrate the balance between exchange such as international conferences. Lastly, consumer protection and innovation, and to help the the WB itself explores the use of new technologies for private sector to develop better sustainable pro-poor development, for example with its Blockchain Lab, an business models; both are core activities of CGAP. Its incubator for learning, experimenting and knowledge core activities also include guidance to policy makers, sharing on Distributed Ledger Technologies. global standard setters and (supra)national regulators, The World Bank Group directly supports the who benefit from CGAP’s research as well as from private sector through the International Finance the projects it conducts with private sector players and Corporation (IFC). The IFC invests in insurance the lessons they provide. All this work also benefits companies and insurance intermediaries in emerging donors and investors. In the area of insurance, CGAP 5. RECOMMENDATIONS 31 works with the Access to Insurance Initiative and the and Asia as a tool to address innovative propositions International Association of Insurance Supervisors to in the insurance space. There is increasing guidance advance inclusive insurance, and is helping regulators for regulators in emerging markets and developing to respond appropriately to m-insurance. An example countries to help them make sense of InsureTech of CGAP’s action research to test solutions based on before being confronted with the need to regulate and behavioral insights and human-centered design is its supervise it appropriately. New ways to use technology cooperation with MicroEnsure to leverage mobile in regulation and supervision are making “RegTech” phones for the delivery of insurance49. and “SupTech” an ever more widespread reality. The World Bank Group launched a Blockchain and AI Lab InsureTech is evolving rapidly. Since 2016, startups to help build awareness and expertise in applications have failed or reinvented themselves and their business of these disruptive technologies and solve development model, and investments have diversified more beyond problems in our client countries, and is exploring the use the United States. Incumbents have reasserted their of artificial intelligence for new solutions to complex role, and much of what was announced as disruptive problems and the effective design development has been incorporated in the traditional insurance policies. The World Bank Group is also working with value chain, which is becoming less traditional in Standard Setting Bodies and the G20 to develop and the process. Prominent recent cases of improper and adapt guidance for policymakers and regulators to unauthorized access to and use of personal data both more effectively support the use of technology in in developed and in developing countries indicate that improving financial inclusion, for example through this topic in particular deserves analysis beyond the the publication of the report called Digital Financial scope of this note, as do the algorithms applied to such Inclusion - Emerging Policy Approaches; and another data. Cybersecurity especially in the financial sector is one on Distributed Ledger Technology and Blockchain. receiving due attention, witnessed by recent publications While this note provides only a snapshot, it documents for example from the World Bank Group. Regulatory a stage in the progression towards ever more actionable sandboxes are increasingly discussed also in Africa recommendations. Contact us to find out more! 32 HOW TECHNOLOGY CAN MAKE INSURANCE MORE INCLUSIVE Endnotes 1. The number of smartphone users is forecast to grow from 1.5 billion in 2014 to around 2.9 billion in 2020 (https://www.statista.com/statistics/330695/number- of-smartphone-users-worldwide/) 2. Products that recognize lower than average life expectancy and provide accordingly higher annuity payments 3. Proprietary global insurance consumer survey in 12 countries on perceptions about technology 4. https://www.linkedin.com/pulse/2117-insurance-odyssey-inga-beale?trk=v- feed&lipi=urn%3Ali%3Apage%3Ad_flagship3_feed%3Bs7JjQfFFJvRPjDTb bWSoEQ%3D%3D 5. http://www.the-digital-insurer.com/wp-content/uploads/2014/10/372- evolution-revolution-how-insurers-stay-relevant-digital-world.pdf 6. http://www3.ambest.com/review/article/January2017/60_AgentsofChange. pdf 7. Opening remarks at Fintech and Financial Inclusion Roundtable in Amsterdam, May 12, 2017 8. Commonly defined as premium to GDP 9. Source: AXCO Insurance Services Global Statistics (2015 figures ) 10. http://data.imf.org/?sk=E5DCAB7E-A5CA-4892-A6EA-598B5463A34C&s Id=1390030341854 11. https://www.insurancefraudbureau.org/insurance-fraud/crash-for-cash/ 12. http://inspool.com/boyracers/ 13. 2016 Swiss Re “Mutual insurance in the 21st century: back to the future?”, which includes a good overview of P2P schemes on p.37 14. http://www.naic.org/cipr_topics/topic_p2p_insurance.htm 15. World Bank Fintech Note #1 (forthcoming) ENDNOTES 33 16. h t t p : / / w w w . e c o n o m i s t . c o m / n e w s / 28. http://www.safeshareinsurance.com/ leaders/21677198-technology-behind-bitcoin- 29. See e.g. http://www.emeraldinsight.com/doi/ could-transform-how-economy-works-trust- pdfplus/10.1016/S1571-5027(07)20011-7, or machine mention in Prahalad’s “The Fortune at the Bottom 17. See for example http://blog.stratumn.com/ of the Pyramid” unveiling-the-lenderbot/ 30. An exception to this has been short term travel 18. See for example http://www.insurancejournal. insurance, which in several cases could be com/news/international/2017/02/06/440629.htm purchased from vending machines on airports 19. http://consuelo.mx/index.html#, http://www. 31. Despite this experience, several insurers launched diariobitcoin.com/index.php/2016/04/25/ scratch card-based microinsurance in Indonesia consuelo-un-microseguro-blockchain-para- starting in 2014, with little success so far trabajadores-migrantes/ 32. The slogan of Cover (http://www.usecover. 20. http://fintank.net/2016/06/09/insureth-smart- com/) contracts/, http://insureth.mkvd.net/ 33. See e.g. http://www3.asiainsurancereview.com/ 21. http://www.safeshareinsurance.com/insurance- News/View-NewsLetter-Article/id/33327/Type/ post-why-unhackable-blockchain-could- eDaily?utm_source/Edaily-News-Letter/utm_ revolutionise-the-insurance-industry/ medium/Group-Email/utm_campaign/Edaily- NewsLetter 22. Morgan Stanley/BCG Global Consumer Survey 2014, BCG e-intensity index, Morgan Stanley 34. Whether to grab attention or to gain competitive Research. advantage via more granular pricing that attracts better risks 23. See e.g. http://www.the-digital-insurer.com/ wp-content/uploads/2014/10/372-evolution- 35. https://www.ft.com/content/7b1226b0-0853- revolution-how-insurers-stay-relevant-digital- 11e4-9afc-00144feab7de world.pdf 36. By contrast, transparent behavior-determined 24. This is why regulators impose minimum premium insurance premium via bonus-malus systems tariffs for the most competitive lines of insurance in motor liability insurance can contribute business in some countries considerably to road safety 25. https://insurify.com/ 37. For an overview of AI applications expected to transform other industries see e.g. https://www. 26. Examples of the use of smartphone cameras cbinsights.com/blog/artificial-intelligence-top- beyond Concierge Distribution include some startups/?utm_source=CB+Insights+Newslette InsureTech startups’ proposition of video- r&utm_campaign=47bc29cda7-Top_Research_ based assessment of car accident claims via Briefs_1_14_2017&utm_medium=email&utm_ smartphones, with the objective of reducing the term=0_9dc0513989-47bc29cda7-88045581 need for costly on-site expertise. A comparable venture aimed at improving the insurability of 38. See e.g. https://www.theguardian.com/ small scale agriculture is discussed in Section 3.5 technology/2017/jan/05/japanese-company- replaces-office-workers-artificial-intelligence-ai- 27. For example 40 million persons in Asia (source: fukoku-mutual-life-insurance Similar coverage http://www.microinsurancenetwork.org/groups/ for banks: https://www.ft.com/content/3da058a0- mobile-microinsurance-covers-40-million- e268-11e6-8405-9e5580d6e5fb people-asia) 34 HOW TECHNOLOGY CAN MAKE INSURANCE MORE INCLUSIVE 39. See e.g. http://www.europarl.europa. 44. https://heyguevara.com/ eu/sides/getDoc.do?pubRef=-//EP// 45. https://www.cbinsights.com/blog/munich-re- N O N S G M L % 2 B C O M PA R L % 2 B P E - startup-partnerships/ 582.443%2B01%2BDOC%2BPDF%2BV0//EN 46. See e.g. http://www.cgap.org/blog/m-insurance- 40. See e.g. https://www2.deloitte.com/ie/en/pages/ ensuring-take-while-doing-no-harm financial-services/articles/RegTech-is-the-new- FinTech.html, https://www.ft.com/content/ 47. See e.g. http://disrupt-africa.com/2017/01/ fd80ac50-7383-11e6-bf48-b372cdb1043a tanzanias-jamii-raises-750k-funding-for- expansion/ 41. www.gartner.com 2013, as quoted in “Insurance and Technology - Evolution and Revolution in a 48. An example is the Indian Jan Dhan Yojana Digital World” (see footnote 5 for source) program, which provides personal accident insurance for active accountholders 42. Insurance and Technology - Evolution and Revolution in a Digital World (see footnote 5 for 49. An example is provided in http://www. source) cgap.org/blog/increasing-immediate-value- microinsurance-poor 43. Quoted from a US insurance commissioner ENDNOTES 35 CHAPTER TITLE 37