Actuarial pricing, capital modelling and reserving

Pricing Squad


Issue 35 -- June 2020

Welcome back to Pricing Squad

Pricing Squad is the newsletter for fellow pricing practitioners and actuaries in general insurance.

Today's issue is about mathematicians and models.


Mathematicians, models and old poets

Recent months have been intense. We are being subjected of extraordinary legal measures and social engineering. The degree to which data, computer modelling and predictions took the centre stage in these crises is unprecedented.

As actuaries and data professional, you must have quickly noticed some glitches in the matrix of which there were plenty. The different ways COVID cases are classified in different countries, the stubborn reporting of number of cases in the absence of statistically unbiased testing, measuring mortality in relation to an unknown denominator.

Here are a few more points I spotted.

Pandemic models suck (not just COVID)

Computer models are intimidating to those who are unfamiliar and maybe this is why politicians trust them. But I am sure you know how dumb they often are (models, not politicians, o wait, actually both).

UK government heavily relied on COVID models by one Ferguson, Neil. The same professor who's modelling:

  • Was responsible for slaughter of 6 million animals during the foot and mouth outbreak in 2001, which left rural Britain economically devastated.
  • In 2002, Ferguson predicted up to 50,000 people would likely die from mad cow disease. In the UK, there have only been 177 deaths from BSE.
  • In 2005, Ferguson said that up to 200 million people could be killed from bird flu.
  • In 2009 Ferguson's advice, said a 'reasonable worst-case scenario' was that the disease would lead to 65,000 UK deaths. In the end swine flu killed 457 people in the UK.

This time around, Ferguson's models predicted 600,000 deaths in the UK and 2.2 million in the USA.

See www.spectator.co.uk or www.cato.org.

Fake data

Another cautionary tale for data professionals and science enthusiasts is provided by the once prestigious "Lancet".

This New England journal has just published an influential paper based on unreliable data and needs to retract. The paper was on effectiveness of malaria-related drugs in COVID treatment.

All data underlying the analysis was claimed to have been collected from multiple hospitals by a private company Surgisphere and provided to the scientist who blindly run with it and published research on this basis.

Further investigation by inquisitive dissenters proved the data not verifiable and some sources claim that it was simply forged by Surgisphere. Surgisphere's employee listed as a science editor appears to be a science fiction author and fantasy artist whose professional profile suggests writing is her fulltime job.

Consequently, authors distanced themselves from their own work and Lancet retracted the paper.

Surgisphere's motivation. Money? Political meddling? Subversion? Who knows but does not look good.

See www.statnews.com.

Other models that suck

If you ever built Solvency 2 or other capital models, you know that Byzantine models producing sci-fi grade results are not just the domain of pandemic modellers.

For those who have not, here are some examples from yet another field, less obscure than financial modelling, i.e. climate change.

On 9 December 2013 The Guardian, quoting US Naval Postgraduate School's Department of Oceanography's complex computer techniques that make its projections more accurate than others, said that Arctic will be ice free in the summer by 2016.

On 24 July 2013 The Guardian, reporting work of professor Peter Wadhams said that Arctic will be ice free by 2015.

Al Gore said the same on 4 December 2009 in USA Today.

In 1972 there was going to be a new ice age by 2017.

And so on.

For countless examples of failed climate models, see www.cei.org.

Why?

Because mathematicians relish abstraction. Once we are given any data, assumptions or axioms, we start building. We are equally happy to work with solid engineering data and to develop a completely abstract non-commutative probability theory. We kind of don't care about the empirical. But we should or we become agents of chaos.

No one captures our nature better that Jan Kochanowski, a XVI century Polish poet. Here is what he has to say about mathematicians.

To Mathematician,
by Jan Kochanowski AD 1583

He discovered the age of the sun and he knows
Just why the wrong or the right wind blows.
He has looked at each nook of the ocean's floor
But he doesn't see that his woman's a whore.

I think it suits Neil Ferguson perfectly.


Brain puzzle

The chart shows non Covid-19 weekly respiratory diseases deaths in the UK. A large chunk of these deaths is "normal" flu.

So here is a puzzle. Let me know if you can solve it.

  1. Why is the count not dropping from 20 March when the lockdown started, relative to prior years?
  2. If the lockdown is not effecting these deaths, how do we know it is reducing COVID deaths?

See www.ons.gov.uk


10 Downing Street Home Insurance Price Index

From 2020 we are sharing with you the "10 Downing Street Home Insurance Index". This is the simplest insurance price index in the world, and provides the average of the top three aggregator quotes for a single property. Have fun!

Conclusion - lockdowns push home insurance prices up.

Increased risk of accidental damage by the PM during a lockdown?


Copyright © 2020 Jan Iwanik, All rights reserved. You are receiving this email because you subscribed to updates from www.iwanik.co.uk. We publish data and analysis for informational and educational purposes only. You can unsubscribe from this list by emailing us.