Restoring innovation to catastrophe modelling

If you’re like me you’ll have to make a conscious effort to achieve focus.

The industry’s innovative past

In the early days of catastrophe modelling, industry leaders pushed the possibilities of risk management using contemporary science, computing, and risk theory. I saw the tail end of this period, and I loved it.

Every year in that period yielded new systems, methods, improvements.

Losing sight of the goal

For decades the industry has not improved methodologies or technology in any fundamental way. We could cherry-pick exceptions and those would be few.

Innovation and risk quantification has given way to production of tertiary outputs that have little to do with risk quantification. We’re better at producing the same outputs, but they remain the same as 15+ years ago.

Is slowing innovation inevitable? Possibly - it makes intuitive sense, but we continue to experience a leaps forward in other industries, especially technology and computing - so the the evidence does not back this up.

Stagnation is a huge weakness for catastrophe modelling, as a function that made its name on innovation - we cannot become stagnant or we lose the point of the function entirely.

We’ve experienced a shift from innovation to stagnation.

Firms will often be satisfied running legacy tools and models if they are good enough - for several years now we’ve been at the point where any insurer can gain access to the same models as established playersi. It’s actually great news for insurance, removing the barrier to entry for new players. But there is still no shortage of flaws to be fixed, new risks that need attention, and new systems to be developed.

So why does it seem like we have stagnated?

The industry is consistently producing non-core outputs

Commoditized risk management

Modelling is outsourced to a few firms - which as above means any insurer can access catastrophe models - but also means there is little variation and limited scope to innovate. Insurers have lost their capability to understand the risk themselves in the first place by relying on a few established modelling firms.

Modelling tweaks not improvements

Similarly, the modelling firms make minor tweaks that barely move the dial. There is a cult around these once innovative firms; questioning their approach is implicitly discouraged lest it harm the credibility of the catastrophe models. Most people outside the risk modelling field can see it, yet I think it is allowed to continue through malaise.

Event responses

Estimating event losses immediately after an event has occurred.

It’s reasonable to expect any company to know how real events affect its’ financial position, but entire cottage industries have been built solely around this practice to the detriment of modelling the risks up front.

This is a process that inherently provides no risk management benefit, as the event has already happened. Empirical evidence from real events is important - but this is different than the futile exercise of estimating instantly after they happen. If it cannot change the outcome then it isn’t risk management.

Regulation

Necessary - but it has hindered innovation. The London insurance market is the most regulated in the world, but this regulation demands firms conduct risk management in a certain way, which can stifle innovation. Inevitably firms concentrate on meeting regulatory demands as a priority rather than improving in other areas.

Meeting regulations - for the most part - should be viewed as baked into every day workflows rather than an add-on or separate workstream.

Other reasons

Elitism

There is often a mindset that certain degrees are a requirement for certain roles, but historic qualifications mean very little when it comes to innovative analytics. In fact I believe this mindset could actually slow teams down.

The roles held by researchers and managers tend to be the ones offered free business class transatlantic travel to Miami and other destinations. No wonder there is so little challenge to the modelling firms and by consquence so little innovation overall.

Failure to train the next generation

Managers who are not quant modellers themselves or whose technical and computing skills have atrophied are not capable of training and leading the next generation.

We therefore have a problem with defining who to recruit, attracting them and retaining them. Time and again good people are lost to tech, banking, and other industries or just don’t apply at all.

Steve Jobs’ lesson

Steve Jobs understood this better than anyone. When he returned to Apple in 1997, the company was stagnant; with dozens of confusing products none of which were innovative or interesting.

Apple went from near-bankruptcy to the world’s most valuable company with this simple principle:

Define your product, then say no to everything else.

Real world challenges

Of course Steve Jobs was a CEO and founder of Apple; he found it easier than most to say no.

In the real world regulatory requirements need to be met, investors and managers need reports. The question isn’t whether to do these things, but how to do them without losing sight of the core product. To achieve this, all waste needs to be ruthlessly eliminated.

The Way Forward

Articulate what the target product of the team is - and ideally what it is not.

We can see what happens to products, firms, teams when this is not defined. The product line loses focus, customer experience deteriorates. It’s no different for any team or industry.

Bad example

Good example

The clarity of the message is important, every day modellers should know the purpose of the team and be empowered to say no to non-core tasks. Over time this leads to increased rate of innovation through gradually eliminating superfluous tasks and focusing on the core vision.

Focus is about saying no.