Rade Musulin Chief Operating Officer, Aon Benfield Analytics Asia Pacific. Analytics to make the Industry Stronger

Transcription

1

2 Rade Musulin Chief Operating Officer, Aon Benfield Analytics Asia Pacific Analytics to make the Industry Stronger

3 Agenda Inspired by the Conference theme of 21 years, I will relate some experiences I have had since 1978 (when I turned 21) which I hope are relevant to Analytics in 2012: o Early years o Technological shifts o Communications o Accurate vs. informative information o A flock of Black Swans allowing for extreme events Keys to successful use of Analytics 3

4 Early Years Lessons from my life at 21 years old 4

5 My life in : o I was studying Chinese history o I had no idea what an actuary was o I used math classes to get my GPA up o I elected to get a math degree when I could not afford graduate school o I had a very diverse background 2012: o Kids usually decide to become an actuary in Year 11 o University is crammed with math o Rounding out process needs to occur with employers 5

6 My first actuarial job 6

7 Job (in)security 7

8 Lessons of the 1970s One never knows where life will lead A diverse background can be helpful we need to help our young employees avoid over-specialization A key survival skill is adaptation Do not fear change that spreadsheet will not put you out of a job 8

9 Technological shifts Surviving the tsunami without drowning 9

10 Technology has dramatically transformed analytics 10

11 For your next geek party 10 trillion digits of π 11

12 Implications Massive databases and computer technology have vastly transformed analytics Our modeling capabilities are incredibly sophisticated, but o Are we aware of how our new technology can affect society? o How do we communicate to stakeholders? o How will public policy planners react? Do we run a risk of being fooled by our models? 12

13 Florida case study Black Box Models 43.7% 43.6% 50.7% M a n a t e e -16.8% 1.9% -6.5% S a r a s o t a -7.9% 8.0% 19.6% 32.1% 9.5% 32.3% H -0.2% a r d e e D e S o t o 2.0% 14.9% C h a r l o t t e -7.7% 52.4% 54.7% 35.9% 4.3% 87.1% 4.7% L e e 24.4% 12.2% H i g h l a n d s C o l l i e r 14.6% 12.1% 0.8% 104.2% 39.0% 23.7% G l a d e s 21.3% H e n d r y 3.0% O k e e c h o b e e 25.5% -5.1% S t. L u c i e -18.3% M a r t i n P -15.1% a l m B e a c h -19.3% B r o w a r d M o n r o e -26.5% M i a m i - D a d e -23.7% 13

14 A crisis born of our technology In Florida we saw a major event expose serious pricing problems in the market This coincided with the introduction of new technology (cat models, Geocoding) that allowed insurers to forensically measure risk Market dislocations and affordability issues arose Politicians intervened Analytics people struggled to understand the implications of their new technology Are we seeing echoes of this in Australia following the recent floods? 14

15 Communications How do we translate gigabytes into sound bites? 15

16 Public policy and the 4 th estate 16

17 Communications failure 17

18 My last speech in New Orleans 18

19 Communications lessons Communication is critically important to analytics A key job of analysts is to translate complexity into clarity Understanding the audience is critical o Management o Financial markets o Government o Media o Consumers Sometimes does not equal 4 Sometimes what is right is wrong All the analysis in the world is of little use unless people act on it 19

20 Accurate vs. informative Are we generating highly accurate wrong answers? 20

21 Spurious accuracy 21

22 Watch the scaling 22

23 Accuracy Numbers you create can haunt you for years Beware of instilling a false sense of accuracy in your estimates Excel can be dangerous o You can compute things to many decimal places o Automatic scaling can create the illusion of an effect If you torture the data long enough it will confess! 23

24 Flocks of Black Swans Why are all our sophisticated tools missing them? 24

25 Extreme events Surprising events seem to be occurring with increasing frequency No one could have seen that coming says the risk manager In hindsight the risks appear obvious Consider two megatrends: o Technological change is transforming events that would have had little effect 100 years ago into major financial catastrophes o The paradox of plenty : modern databases and models offer a sophisticated picture of risk, tempting us to believe that just because a model is complicated it must therefore be comprehensive 25

26 Eyjafjallajokull Iceland is one of the most volcanically active places on earth Since the settlement of Iceland in 874 AD there have been at least 13 major eruptions, and over the past 500 years fully one-third of the global lava output has occurred there Risks to airlines from volcanic ash have been well known for decades There were no eruptions in Iceland since the advent of jets Airlines appeared to be caught off guard in

27 Supply chain problems - Tohoku Japan earthquake Tsunami Earthquake models did not account for tsunami Contingent business interruption exposures were understood, but not acted upon by many Political risk was not factored in Massive economic risks were discovered from a chain of events that, in hindsight, should have been apparent Were we lulled to sleep by our technology? Parts shortages in the US Nuclear meltdown 27

28 Queensland floods Human activity is creating change that is affecting insurance losses o Flood control (or lack thereof) o Deforestation o Pumping aquifers and sinkholes o Cities in river deltas and subsidence Queensland dams were built following massive flooding decades ago Mission was expanded to include drought control When ENSO conditions combined with IOD in 2011 flooding occurred when dams were full; releases led to flooding 28

29 Thailand floods Thailand has suffered from major floods for hundreds of years The government has invested in flood control defenses Global (Japanese) firms built large factories and used rivers for transport Catastrophe models did not cover Thai flood; it was off the radar Risk managers did not see the exposure Massive losses resulted; was this an ERM failure? 29

30 Lessons from recent black swans Supplement traditional loss modeling with carefully constructed scenario analysis rooted in the historical record yet reflecting demographic and technological change: o The long term historical record of the type of events which could trigger insured losses, looking beyond recent insurance loss statistics o Demographic changes, particularly involving the construction of large cities in locations historically prone to earthquakes, floods, and tropical cyclones o Technological advances, such as the introduction of just in time manufacturing o Economic interdependency, particularly surrounding supply chains o Effects of human activity on historical loss exposure, such as how new fire control strategies in forests might affect wildfire potential o Human decision exposures, such as standards governing water releases from dams Avoid the paradox of plenty, where the quality of overall risk assessment can actually degrade despite better analytical tools 30

31 Concluding thoughts Analytics to make the Industry Stronger 31

32 Our theme Faster more likely to crash? Higher farther to fall? Stronger are we more secure? 32

33 Take aways The push to lock in careers at and early age and to specialize education means employers have to focus on completing the education of their analytical colleagues Technological change can have unintended consequences Analytics must be viewed from many perspectives, including business, academic, public policy, and the media We must guard against the illusion of spurious accuracy, of believing our tools are better than they are We must do a better job of dealing with black swans : o Pay attention to history o Form multidisciplinary teams o Use scenario testing o Allow for the human factor o Understand the limitations of your models! 33

34 Speaker Contact Information Rade Musulin Chief Operating Officer, Aon Benfield Analytics Asia Pacific Aon Benfield Level 29, 201 Kent Street Sydney NSW 2000 Australia rade.musulin@aonbenfield.com This presentation has been prepared for the FSAA National Conference May Opinions put forward herein are not necessarily those of FSAA and FSAA is not responsible for those opinions. The information presented at the seminar is of a general nature and a reader of this presentation must seek their own independent advice before using it for any purpose. 34