November 19, 2016

Future Scenarios – Insurance Goes Insane


This is a ‘Future Scenarios’ post where we break down a conceptual future scenario and ask ‘what if.’  The idea is to challenge the political narratives surrounding economic development, whereby all forecasts are built on constant, linear growth, and the continuous expansion of the industrial economy without any real thought to emergent social, cultural and environmental constraints. By challenging these narratives that exist, Future Scenarios emerge that shed light on new models for growth and economic development. 

What if Insurance Goes Insane?

Insurance is based on risk, risk that is calculated using linear models developed by actuaries. We live in a non-linear world where risk happens not at all, then all at once. Think about the insurer AIG pre-2008. It was a gem of the market and doing more than 80 Billion USD in revenues in 2006 and 2007, and then bang:

“The AIGFP division ended up incurring about $25 billion in losses, causing a drastic hit to the parent company’s stock price.” A Case Study of AIG

The subprime mortgage crisis blew a hole in the company’s balance sheet; consequently, they nearly took down the global economy and needed a $182B bailout by the US Federal Reserve. The irony is that none of AIG’s executives even saw a risk to profitability, let alone a $25B hole in the balance sheet:

“Over the course of the day, 14 AIG executives made presentations … Every one of them said the same thing: there was little or no chance that the tranches that AIG had either insured (in the case of FP) or bought (in the case of other AIG divisions) could ever lose money.” All the Devils Are Here


Marotta on Money

Their models showed no red flag.  It wasn’t a problem, until it became a problem. Then risk went exponential, their models blew out, and suddenly the world was in the throngs of a Credit Crisis that nearly derailed the global economy. Everybody is aware of what happened with Lehman in ’08, but it was AIG and the “reckless issuance of nearly half a trillion dollars in toxic credit-default swaps” that nearly brought the whole world down.

While there is no real probability of a subprime mortgage meltdown right now, the net risks across all sectors – social, cultural, economic, environmental, etc – have increased 100X  in the last decade and most insurers are still using the same linear models that rely on historical data and symmetrical returns to calculate risk in an asymmetrical world.

But amidst the doom and gloom of global risk is a shining beacon, which is why in this edition of Future Scenarios, we look at the insanity of insurance in the post-industrial era and start to imagine what we can expect when big risk starts to roll.


Cyborgs and Earthquakes

Swiss Re is one of the largest global reinsurers in the world. In its latest SONAR report, the company identifies the top three risks it faces in the current macro environment: emerging markets, monetary policy, and a fragmented internet.

All pretty standard. It then goes on to list ’emerging risks’ that are anything but conventional:

“Legal and pricing risks of the sharing economy. Swiss Re says traditional insurers could face pressure from new players (peer-to-peer insurers) that can reach new clients through a sharing economy situation. As well, insufficient loss experience could expose insurers to inadequate pricing models for hybrid risks in the sharing economy.

Crisis of trust. As citizens increasingly distrust governments, large corporations and traditional media, this could spill over to a distrust of insurance, which could hurt business. If customers are more hostile, then unwarranted claims could spike.

Mass migration. Global migration has become a concern due to instability in many places around the world. Swiss Re said it could have relatively minor impacts on insurers, such as the need to reassess risk exposure of buildings due to higher-than-expected occupancy from refugees.” Swiss Re SONAR report


It goes on to list:

“viral leaderless mobilization, the future of work, precision medicine, neutraceuticals, gene drives, human cyborgs, beef, fintech risks, blockchain risks, phoney data, human induced earthquakes, and ocean pollution from microplastics. They are just beginning to scratch the surface.”  Swiss Re SONAR report

In an era of digital disruption, geopolitical tension, class struggles, climate turbulence, and job automation, who knows where this all leads. Not that we should feel sorry for insurers, they have had a pretty nice run over the last few decades:


Insurers break their premium growth down into ‘life‘ and ‘non-life‘ and ‘advanced’ and ‘non-advanced‘ markets. They drive revenues from premiums, and their profitability is determined both by how much they pay out in claims and how much interest they earn from assets. Simply on the basis of extreme monetary policy, the core insurance business model is threatened:

“Insurers globally are having to come to terms with the idea of “lower for longer” interest rates, making deep changes to business models that had been unaltered for decades. Whereas previously they might have clung to the hope that higher rates were around the corner, there is a realisation that the industry has to do things differently — from investing in assets that might once have been seen as too risky, to experimenting with new products.” Insurers: Forced to Dig Deep

Add in the emergent environment of radical risk and the industry dinosaurs may soon be on the verge of extinction.

But that’s only half the problem. If insurers have no clue how to actually predict + manage risk in the emergent environment, and a catastrophe in one form or another happens, then will another AIG be waiting in the wings for a state bailout?

How do we predict + manage risk in a rapidly changing environment marred by continuous chaos and uncertainty?

Chaos Theory >Radical Risk Scenarios

“Chaos theory is the field of study in mathematics that studies the behavior and condition of dynamical systems that are highly sensitive to initial conditions—a response popularly referred to as the butterfly effect.”Wikipedia

The butterfly effect is when ‘small causes can have large affects’ leading to scenarios where future behavior is fully determined by initial conditions. The result is chaos (from a physics perspective).

The name, coined by Edward Lorenz for the effect which had been known long before, is derived from the metaphorical example of the details of a hurricane (exact time of formation, exact path taken) being influenced by minor perturbations such as the flapping of the wings of a distant butterfly several weeks earlier.” Wikipedia


Last year, two scenarios would have been considered radical going into 2016 by the mainstream:

  • Brexit
  • Trump

They both happened. The polls failed to predict them. Both were such longshot odds on their own, let alone both happening simultaneously in the same year. And none of the dire consequences predicted by those same institutions if Brexit or Trump happened actually came true. The mainstream was so wrong about both the probabilities and the expected outcomes that you would be loath to give their forecasting any credibility in the future.

This doesn’t bode well for 2017 and beyond. With major elections in the EU in 2017, populist uprisings globally, rapidly increasing climate risk, shaky social foundations, cultural tensions and whatever else you want to add into that mix, what happens if we all these risk factors scale at once to create a scenario that is uninsurable:

  • Climate catastrophes
  • Loss of life
  • Mass Unemployment
  • Food shortages

What’s the solution?

Machine Learning and Community

The key to insurance is pricing the risk. If you have no real way to measure the actual risk, the entire model can be blown out in a scenario of radical risk:

 “We are trying to put a price on an event that has not happened yet. No matter what models you use, that is fundamentally true.Intelligent Insurer, April 2016

Then once the risk is priced, you have to pay out the claims. But because of the business model of insurers, they are incentivized not to pay out claims:

“If you tried to create a system that brings out the worst in people, you would end up with one that looks like the insurance industry.” The Science Behind Lemonade

The solution involves rectifying these two core problems.

Machine Learning

“Machine Learning focuses on the development of computer programs that can teach themselves to grow and change when exposed to new data.”


As the data comes in, computers learn and automatically adjust risk forecasts and pricing models. Generally, it is assumed that there is a 3-5 year product development curve in insurance, but with machine learning, this curve will be dramatically reduced:

“Typically, it takes more than a year to launch any kind of new technology, and takes 3-5 years of data to train underwriting models. Meanwhile, more data is available now than ever before — and it’s available in real time. This lag of product development represents the opportunity for startups!” The Next Fintech Frontier Insurance

Brian Duperreault, an insurance industry legend, came out of retirement in 2013 to start Hamilton Insurance Group. Duperreault was approached by the Two Sigma hedge fund that uses machine learning to create investment strategies:

“Two Sigma Investments is a New York-based hedge fund that uses a variety of forms of technology including artificial intelligence, machine learning and distributed computing to inform and drive investment strategies. In a nutshell, it tries to remove the potential for human emotion and error from investment decisions, and it has been very successful employing this strategy so far.” Intelligent Insurer, April 2016

Hamilton Re’s suite of insurance products is built using a ‘proprietary risk platform’ based on machine learning.


Machine learning can be used to build the pricing models and adjust them as new data is presented to the computer. The construction of these products would give Hamilton an advantage against incumbents who are “grossly over-expensed and marred by inefficiencies and inadequate use of data.”

There are other enabling insurance models that can be built using this approach. One of Hamilton’s joint ventures in Africa, Blue Marble MicroInsurance  is “a consortium of eight companies collaborating as a for-profit social enterprise to extend protection to the emerging middle class:”

Blue Marble Microinsurance’s founding partners know that the ability to manage and finance risk is critical to the development of society – any society, but most urgently to those struggling to gain a stable toehold in their pursuit of education, jobs and a prosperous future.” Microinsurance Incubator Announced

One of their prototype ventures – the first of ten over ten years  – is drought protection for maize farmers in Zimbabwe, which incorporates technology and data science to measure what’s actually happening in the farmer’s fields:

  • “First, the design of a proprietary index to support the insurance product is a collaborative innovation of data scientists and agronomists from the eight consortium companies” #datascience 
  • “The second feature is the use of two-way mobile communication with customers to improve the overall proposition continuously” #mobile 
  • “Third, the use of innovative point-sensor technology to measure rainfall and plant health throughout the growing season will complement traditional grid remote-sensing data to create a higher-resolution parametric insurance cover.” #sensors  BlueMarble Launches First Venture


There are various points of innovation here in relation to the insurance model:

  • adjustable risk
  • adaptive pricing
  • machine learning

Instead of an actuary developing a model for farmers in Zimbabwe based on historical data and generic stats, and pricing the product based on some spreadsheet calculation, this product is designed with collaboration among partners, live data, machine-learning models, and open dialogue with the farmers themselves. This combination, from a risk perspective, is revolutionary for the insurance industry.


Let’s say that many more insurance products of this nature are launched over the next five years. What if, however, in a radical risk environment, the models get it wrong, resulting in massive claim liabilities. A core part of the insurance business model is to make it as difficult as possible for people to be made whole on their claims.  Insurance companies have created behavioral incentives for consumers to cheat them, and in turn used these cheats to justify onerous and unreasonable procedures for making claims.

+ Lemonade and the Insurance Business Model

This can change with a greater community focus, as we have seen with the launch of Lemonade in New York State:

“homeowners and tenants across the state can get insured and settle claims on the spot, across multiple devices.” P2P Firm Lemonade Launches


Naturally, major global insurers wouldn’t be able to implement this model or they would go bankrupt. Lemonade is able to implement it because of the BMi (Business Model Innovation) around the community:

  • Individual takes out an insurance claim – in a few minutes – via the Lemonade Bot
  • Lemonade takes 20% as a fee, the remaining 80% is put into a Pool based on the Cause the individual chooses
  • ‘Cause Pools’ aggregate the remaining 80% from all individuals into one common ‘shared risk’ account
  • Claims – which take only minutes to successfully file – are redeemed and paid out from the Cause Pool, not from Lemonade
  • Reinsurers cover bigger risks that cannot be covered by Cause Pools

Lemonade is applying the behavioral science of Dan Ariely to create an insurance model that becomes “a social good rather than a necessary evil:”

BlueMarble Micro is focused on bringing machine learning to create better models and risk metrics for farmers in Zimbabwe, whereas Lemonade is focused on using the community of homeowners and tenants to create new behavioral incentives. But both are getting to the core in an effort to create ‘social good’ and profit simultaneously.  One looks more like an insurance company and one looks more like a tech company, but both companies have elements of machine learning and community embedded in their model with the hopes of completely reinventing risk measurement and reward.

Blockchain Future

What would a discussion of future insurance scenarios be without blockchain, described by The Economist as “the great chain of being sure about things:”

“The embers of innovation are beginning to char the massive $1.2 trillion underbelly of the largest industry in the world.” Blockchain is empowering the future of insurance

In the future, the greatest way to protect against radical risk in the insurance industry will be to distribute the risk across society, rather than relying on big institutions to manage that risk themselves:

“P2P insurance empowers policyholders to a greater portion of the premiums rather than the individual private wealth managers working to produce returns for insurance companies.” Blockchain is empowering the future of insurance

P2P (peer-to-peer) is part of the ‘trust wave’ taking place in the sharing economy, but it can’t come to fruition without innovation such as blockchain, which creates an irrefutable, unchangeable record that something did or didn’t happen.


+ Trust Mechanisms in the Digital Age

One of the advantages for insurers, compared to banks, is the relatively high degree of trust millennials place in insurance companies. About 61% of millennials in the US trust insurance companies to ‘do what’s right for their customers.‘ Furthermore, there is a huge regulatory component to insurance, so it is very difficult for a Silicon Valley startup to come and up-end global insurers in the way Uber tipped the taxi industry:

“Insurance is one of the most difficult businesses to start, because, well, the regulators don’t actually want new players in the market The reason for this is risk — insurance is all about having a strong system and balance sheet to manage risk. However, as a result of its difficulty, insurance is now behind the times in terms of technology. This leads to a perfect storm to disrupt the industry.” How to win in the next fintech frontier – insurance

The ‘disruption’ itself will need to come through new models of risk pricing, claims management, and technological implementation such as machine learning, artificial intelligence, or blockchain.  It’s hard to predict which firms will lead the new frontier, but no matter who wins the long game, it will require a multi-faceted approach beyond simply technology.

The key for the entire industry, going forward, is trust. In a future scenario where insurance goes insane, society needs to be able to trust that insurance companies will remain solvent and bankable enough to payout claims.  In an emergent environment of radical risk where anything can happen, we need insurance companies to step away from political narratives and step up to become stewards of risk for society. As the peer-to-peer environment around the sharing economy and blockchain evolves, we can build new models to distribute that risk across society and return insurance back to its village, pre-Industrial roots.