Black Swan events cannot be predicted. Who could have predicted that Lending Club would do something wrong that caused the Founder CEO to leave and the stock to tank? We cannot plan for hurricanes, tsunamis and other extreme weather – we just put that in the Force Majeure clause. Yet big companies have to plan for these types of eventualities and consumers insure against them – making the subject important for the Global 2000 and mission critical for Insurance companies. Insurance companies do have to predict what has historically been called unpredictable. Having come off a lousy ski season, I am more inclined to believe in global warming and therefore an increase in extreme weather related risk. Seriously, business people who have anything to do with winter sports, tend to plan seriously for global warming. It is not an academic debate for them. It is certainly not an academic debate for Insurance or Reinsurance companies who have to payout if something goes wrong. It is certainly a serious concern for those investors who searched for yield in a ZIRP world and found the big payouts from Catastrophe Bonds (Cat Bonds). If the incredibly complex system known as the earth’s climate has been disturbed, all our conventional modelling systems will have to be discarded. However the global climate is only one of many massively complex networks where a butterfly flapping its wings can do a lot of damage – financial markets triggering price risk, social networks triggering reputation risk, cybercrime networks triggering operational risks. Traditional statistical models don’t work so well in these chaotic/complex networks. Any companies that offer solutions to this modelling problem could do well. We cover two of them – Praedicat and Meteo Project. Please note, this post will NOT cover either the science or politics of climate change, we restrict our ourselves to the geeky subject of non-linear modelling.
Linear vs Non-Linear Models
Like so many others I read Chaos: Making of a New Science by James Glick in 1987 and it deeply influenced my thinking. I went on to read some books about Complexity Science and what the Santa Fe Institute were doing and around 2002 got involved with a tech startup that applied these theories to the hairball complexity of dependency management in giant old legacy software systems (there was a successful exit). So when the Global Financial Crisis happened in 2008, it simply looked to me like a chaotic/complex system being disturbed. This is the same thinking that one needs to apply to climate and other network driven risks. Disturbed complex systems turn chaotic and that is what leads to risks such as more violent weather, mass law suits, reverse network effects based on reputation risk etc. This is the cheery subject that Insurance/Reinsurance actuaries and Cat Bond investors/traders need to think about.
Praedicat – big data for emerging liability risk
Praedicat is a Los Angeles based venture that has raised $12m (but their last raise was in 2013, so they have either become cash flow positive or they could be having trouble raising money). They focus on “improving the underwriting and management of liability catastrophe risk” using big data analytics. Insurance companies are a natural target, but the market is broader. All big companies work on “enterprise risk management”, the complex intersection of operational risk, market risk, reputation risk and regulatory risk.
If you want to really dig into this, read the Lloyds report on Emerging Liability Risks (Harnessing Big Data Analytics). Or read DailyFintech for a quick heads up (join over 9,000 of your peers and subscribe by email, its free). We read this kind of wonky stuff so that you don’t need to.
Mostly when we read about Big Data Analytics it is about finding that one needle in a haystack that will tip a consumer into hitting the buy button (or some other action that leads to the buy button). Praedicat is using Big Data Analytics in a different way. This is hard core enterprise stuff (which will impact consumers in their premiums). The Lloyds report is co-authored with Praedicat and linked to from their front page. That is some useful validation for a startup.
The Lloyds/Praedicat report focusses on the tipping point that triggers class action lawsuits. That makes sense. It is something that Board Directors worry a lot about and a simple rule of big companies is that whatever Board Directors worry about tends to get budget allocation:
“this approach estimates the probability of a general consensus being reached that exposure to a substance or product causes a particular form of injury. This is the critical threshold at which lawsuits become more likely to succeed; a liability catastrophe could emerge if a successful lawsuit gains traction and triggers mass litigation. This information is then overlaid on an insurer’s portfolio to identify potential accumulations of liability risk. The analysis can be used to develop quantitative estimates of mass litigation, allowing a liability catastrophe model to be built from the bottom up.”
This is not Force Majeure. This is something that BigCo did – or did not do but should have done.
One new factor in this kind of risk is the reputation risk and mass mobilisation of consumer action enabled by mass adoption of social media. We have witnessed corporate reputations go down in flames thanks to one tweet or video that goes viral. That in turn can trigger class action law suits. You cannot get rid of social media or class action law, so the only choice is to model what kind of behaviour triggers that kind of consumer reaction and then figure out how to price that risk. That is what Praedicat is trying to do.
They are less concerned with Force Majeure type events. Companies leave that problem up to Insurance. They are less worried about what Mother Nature might do than what BigCo will do such as “large-scale catastrophes such as bodily injury from toxic chemical exposures or property damage from accidents during energy production.”
Meteo Protect is a Paris based venture is coming at this from a different direction. They are focussed on one type of Force Majeure risks, those arising from climate change. They put a number on the problem, claiming that weather can cause $500 billion in lost profits in America alone. This is not just about catastrophic risk. The low hanging fruit is being able to predict when consumers will buy ice cream vs umbrellas (taking a most simplistic example). Getting that inventory decision right is critical and depends on weather forecasting. So is Meteo Protect a weather forecaster? Not really. The key is in the Protect part of the name. This is really about taking existing weather forecasts and intersecting them with consumer behaviour and your product line. In a White Paper on their site they take the bold position that Credit Rating Agencies should include climate vulnerability into their analysis.
This is the sort of market that big enterprise vendors such as IBM, SAP and Oracle are focussed on as well as the management consulting firms serving the Global 2000. Meteo Protect uses SAP HANA technology. See our earlier coverage of Meteo Protect here.
Unpredictable Reputational Risk
The Lending Club story is at heart a reputational risk issue based on an ethical lapse and a failure of business process management controls. Insurance companies do offer reputational risk insurance. For example, Allianz focusses on these risks:
- Health and safety incidents
- Operational crises and events (e.g. pollution)
- Product recalls and quality control errors
- Business and service interruptions
- Financial losses and irregularities
- Negative associations with third parties Management and governance topic
- Legal and regulatory investigations Allegations over business practices
- Ethical violations and challenges
In the same week that we had the Lending Club meltdown, we also had Salesforce.com suffering an outage lasting many hours causing huge amounts of lost productivity for millions of users. The “sky is falling” crowd rushed to say that lending Marketplaces and Cloud Computing are just hype and passing fads (a bit of an over reaction to put it mildly). Both incidents are likely to trigger class action lawsuits and an increase in Insurance costs. In both cases, people will say “but it was unpredictable”. That answer is no longer acceptable. If we don’t have the tools and processes to predict events like that, then we had better work on them. Ventures that solve that type of problem should do well both with Global 2000 and with Insurance companies.