Session 24 PD, Actuarial Climate Research Update and Practice Applications. Moderator: R. Dale Hall, FSA, CERA, MAAA

Transcription

1 Session 24 PD, Actuarial Climate Research Update and Practice Applications Moderator: R. Dale Hall, FSA, CERA, MAAA Presenters: Sam Gutterman, FSA, CERA, FCA, FCAS, HONFIA, MAAA R. Dale Hall, FSA, CERA, MAAA Ronora E. Stryker, ASA, MAAA SOA Antitrust Disclaimer SOA Presentation Disclaimer

2 Social Discounting SAM GUTTERMAN, FSA, FCAS, MAAA, CERA, FCA, HONFIA Session 24: Actuarial Climate Research Update and Practice Applications October 24, 2016

3 Agenda The project Discounting Social discounting What is it How it relates to climate change Inter-generational equity Traditional economic approach Uncertainty Structure and application Real options Challenges 2

4 The project Funded by the Society of Actuaries It has a recently established Project Oversight Group Currently a work-in-progress 3

5 Discounting Purposes*: Recognize the time value of money Reflect or provide for risk in projected future asset cash flows or Reflect the opportunity cost of holding an asset Rationales** Consumption rate at which society is willing to substitute consumption in the future for consumption today Investment rate at which equates the need to invest less than a dollar today to obtain a dollar of benefits in the future Different bases for discounting are possible In cost-benefit analyses a key element Always a relationship between time and risk/uncertainty * International Actuarial Association (2013) Discount Rates ** Arrow et al. (2013) Determining Benefits and Costs for Future Generations 4

6 Social discounting One application of discounting Market discount rate Often used by actuaries Reflects the preferences of market participants who express their value of time through the use of transaction prices for goods or services Social discount rate A parameter that measures the importance of the welfare of future generations relative to the current generation Observable market prices do not reflect costs and benefits to society, as they only reflect views of the market participants Perspective is the country s policymaker 5

7 Why shouldn t market rates be used? Social discount rate not a market rate because of imperfections of market rates relative to the purpose of this application Market prices (costs or benefits) don t include costs to society Shorter-term focus Irreversible environmental damage Doesn t consider non-financial costs Irrationality of certain market participants Effective hedges unavailable 6

8 Importance In many quantitative assessments over the long-term, the choice of discount rate represents one of the key areas of dispute Discount factor at 200 years: 1.0% % % % Recent survey of 197 climate change economists* Range 0% to 10%, with 92% comfortable between 1% and 4% Mean 2.0%, median 2.25% * Drupp et al. (2015) Discounting Disentangled 7

9 Relation to climate change To assess the current value of costs associated with mitigation of or adaptation to the effects of climate change A great deal of CO 2 emissions remains in the atmosphere for centuries Therefore, very long time frames and multiple generations are involved Almost irreversible A key reason why climate change costs are looked at differently from other long-term costs Other emissions have much shorter half-life, e.g., methane 8

10 Climate change process Greenhouse gas emissions Accumulated greenhouse gases in atmosphere and oceans Mitigation activities Warmer, more volatile weather conditions and other climate change effects Adaptation activities Economic damages - sudden and slow onset Time and risk preferences Estimated value in present value terms 9

11 Inter-generational equity A concern because climate change will affect many generations In addition to the usual issue of intra-generational equity The question: In current decision-making, how to fairly weight the irreversible costs associated with current action expected to be borne by future generations? 10

12 The Ramsey formula Economist Frank Ramsey (1928) presented a social discount rate approach to analysis of savings Commonly used by economists for cross-generational discounting: r = ρ + ƞ g where r = social discount rate ρ = pure rate of time preference ƞ = elasticity of marginal utility (in terms of a utility function) g = per capita growth rate of consumption 11

13 The Ramsey formula (2) In macro-economic terms, the consumption rate of interest the rate at which society would substitute consumption in a future year for consumption in the present First term: pure time preference Viewed by advocates of speedy environmental action (e.g., Lord Nicholas Stern) as an ethical issue shouldn t put the next generation in a worse position compared with the current one Makes the case for a 0% rate Second term: growth factor Represents the expected extent that the future will be better off than the present Doesn t take into account uncertainty 12

14 Uncertainty Any projection of the effects of climate change involves a great deal of uncertainty Future GHG emissions Natural offsets Effects of climate amount, frequency and timing Costs of mitigation Discount rate Environmental decisions should consider uncertainty Classical actuarial theory Either an increase to expected cash flows or reduction in discount rate In this context, typically use a reduction in discount rate Often characterized by a declining set of discount rates, adopted by at least United Kingdom and France in their decision-making Ramsey formula, which assumes certainty, has been adjusted by addition of a third term: 0.5ƞ 2 σ 2 13

15 Structure of social discount rates Practice to date by national governments is split between: Level discount rates Simple Example: U.S. government requires alternative discount rates: 3%, 3.5% and 5.0% Declining discount rates More consistent with currently accepted theory and reflection of uncertainty U.K. start at 3.5% declining to 1.0% after 300 years France start at 4.0% declining to 2.0% after 30 years 14

16 Application of social discount rates Important to recognize who is the user of such costbenefit analyses Ramsey formula often applied to consumption, reflecting society s utility function But some have expressed skepticism regarding the ability to accurately quantify this function Actuaries usually apply discount rates to cash flows or risk-adjusted cash flow equivalents But it is difficult to incorporate non-financial costs, such as human life, residential oceanside property and heritage assets 15

17 Real options Definition The right, but not the obligation, to undertake an initiative, such as deferring, abandoning, expanding, staging, or contracting a capital investment project In valuation, many options exist for a public policy decision-maker for a climate change initiative To act on or defer (procrastinate) action For example, based on New information, future resources availability, or expected (hoped for) new technologies For discussions of climate change prior to 2007 It was common to assume future costs won t be that bad and that future technologies will be able to provide a cost effective response Difficult to quantitatively reflect these options 16

18 Challenges in completing report Coming to an objective conclusion A political issue, as it involves allocation of resources ( fairness ) between, for example Developed and developing countries Well-off and vulnerable who are most affected Unsurprisingly, those who advocate immediate action tend to justify a lower social discount rate, while those who advocate limited or deferred action tend to justify a higher social discount rate Can it equally apply to mitigation and adaptation decisions? An issue in which there is both High degree of uncertainty, speculation and subjectivity Jobs and money at risk 17

19 Thank you for your attention 18

20 Measuring the Social Media Response: An Actuarial Perspective on Catastrophe Loss Control and Prevention R. DALE HALL, FSA, MAAA, CERA, CFA Managing Director, Research Society of

21 Evolution of social media in insurance loss prevention and loss control Social Media becoming an increasing part of the insurance industry communication message Natural extension of messaging for policyholders & public affairs Increasingly important in communication prior to, during and after catastrophes 2

22 Poll: I am active with a personal account on A.Instagram B.Twitter C.Facebook D.Snapchat

23 Poll: I am MOST active with a personal account on A.Instagram B.Twitter C.Facebook D.Snapchat

24 Evolution of social media in insurance loss prevention and loss control Social Media evolution is dynamic Where are your policyholders today? Where will they be tomorrow? Blurring of lines between social interaction and information flow 5

25 Data intensity and its study Actuarial focus on data plenty to be collected through catastrophic events Objective: Use social media as a source of data and information Most common and available information source is Twitter Hypothesis: Each unique cat event has its own distinct Catastrophe Communication Signature (CCS) Evidence and Measurement Early Warning System effectiveness Information flow Loss Control Society of Actuaries 2016 study Tweets from 3 recent catastrophes Observe changes in signatures & outcomes 6

26 Tornados Studied Location Joplin, MO Washington, IL Rochelle, IL Date Sunday May Sunday November Thursday April Touchdown 5:34 PM 10:59 AM 6:40 PM Severity EF-5 EF-4 EF-4 Casualties 158 deaths injuries 8 deaths 190 injuries 2 deaths 30 injuries

27 Methodology Data collected over a two-week period (3 days prior, 11 days after) each event Data sorted for catastrophe relevance Example: TORNADO OR EF OR [contains INJUR*] OR [contains INSUR* ] OR. AND {Cat Location Identifier} Approximately 500, ,000 tweets per event and 70+ pieces of information for each tweet Foreign language filtering to get to final 100, ,000 tweets Groupings of tweets into 5-minute intervals; Main focus in visuals over a 24-hour period 6 hours prior to touchdown 4 hours During Event 14 hours After Event Categorization of tweet volume by key industry groups for comparison 8

28 Observations Growth of Social Media Communications Population of Impacted Metro Area Importance of Early Warning Systems April 2012: Commercial Mobile Alert System (CMAS), interface to the Wireless Emergency Alerts (WEA) service goes live Time of Day / Day of Week Influence Length of Overall Event 9

29 Poll: I currently am registered to receive Wireless Emergency Alerts through a cell phone

30 Joplin 2011 CCS 1,500 Tweet Volume in 5-Minute Intervals Joplin, MO; May 22, ,250 1,000 Tweets

31 Washington 2013 CCS 1,500 Tweet Volume in 5-Minute Intervals Washington, IL; Nov 17, ,250 1,000 Tweets

32 Rochelle 2015 CCS Tweet Volume in 5-Minute Intervals Rochelle, IL; Apr 9, ,500 1,250 1,000 Tweets

33 Industry Loss Control / Prevention Tendencies Today vs. 2011: Strong increase in use of Social Media for Alert / Prevention by Trade Associations and Companies Each company with unique focus Trade Association alerts to industry and consumers Strong growth in followers Use by distribution as communication venue for loss reporting and loss control Companies and Distribution using social media as a source of information flow 14

34 Overall Comparisons 15

35 Future Observations and Research Deeper study of multi-tornado events down to location level Twitter has specific geo-location information for some tweets, but not all What is the future of Twitter? Review of Twitter sentiment ratings as storm evolves, occurs and passes 16

36

37 The Actuaries Climate Index TM Countdown to Launch Ronora Stryker, ASA, MAAA SOA Research Actuary Session 24: Actuarial Climate Research Update and Practice Applications October 24, 2016

38 Agenda Overview of ACI TM and ACRI TM ACI TM Data Sources ACI TM Calculation ACI TM Results Uses of ACI TM and ACRI TM 2

39 Overview ACI TM and ACRI TM 3

40 ACI TM and ACRI TM Structure Development Sponsored By North American Actuarial Orgs: AAA CAS Climate Change Committee and Index Working Group CIA SOA Consultant: Solterra Solutions climate science Building Off Phase 1 Research Determining the Impact of Climate Change on Insurance Risk and the Global Community 4

41 ACI TM and ACRI TM Goals Create an objective index that measures changes in climate over recent decades Inform the insurance industry and the general public on the impact of climate change Create an easy to understand index, but not simplistic Promote the actuarial profession 5

42 ACI TM Basics Measures change in frequency of extreme climate events by comparing recent climate extremes to historical events Express changes as standardized anomalies, e.g., X = (X X ref ) / σ ref (X) = ΔX / σ ref (X) Base/Reference Period:

43 ACI TM Basics Six variables: 1. Temperature Highs 2. Temperature Lows 3. Precipitation 4. Drought 5. Wind 6. Sea Level Data based on measurements from meteorological stations and taken over a 2.5 by 2.5 grid (275km x 275km at equator). All data is publicly available 7

44 ACI TM Basics Initially Canada and U.S. focus Examine countries separately or together 12 subregions Publish index and related information on website Status: Preparing to launch the website 8

45 12 Subregions 9

46 ACRI TM Basics Combine components of ACI with exposure (population is the proxy for exposure) Measure correlation of economic losses by peril to the exposure in determining index Mortality and morbidity vs heat (4) Flood damages vs max. 5-day precipitation (8) Crop damages vs consecutive dry days (3) Wildfire damage vs consecutive dry days (2) Wind damages vs wind power (7) Status: Still in development 10

47 ACI TM Data Sources 11

48 ACI TM Data Sources - Temperature GHCNDEX indices* based on NOAA s Global Historical Climatological Network (GHCN-Daily) otx90 = 90%ile warm days otn90 = 90%ile warm nights otx10 = 10%ile cold days otn10 = 10%ile cold nights * Produced as part of the CLIMDEX project by the Climate Change Research Centre, at The University of New South Wales, Australia. 12

49 ACI TM Data Sources - Precipitation GHCNDEX indices* based on NOAA s Global Historical Climatological Network (GHCN-Daily) o Rx5day = monthly maximum five-day precipitation data o CDD = Max days per year with <1mm precipitation o Drought index = 1 value of CDD/year o Linear interpolation to obtain monthly * Produced as part of the CLIMDEX project by the Climate Change Research Centre, at The University of New South Wales, Australia. 13

50 ACI TM Data Sources - Wind Index derived from NOAA Earth System Research Laboratory data: o Daily mean wind speeds o WP = (1/2)*ρ*w 3 Where ρ is air density, w is daily mean wind speed 14

51 ACI TM Data Sources- Sea Level At tide gauge stations along US and Canada coast o Data provided by Permanent Service for Mean Sea Level (PSMSL), part of the UK s National Oceanography Center o Data matched to grids used for other variables 15

52 ACI TM Calculation 16

53 ACI TM Calculation Steps 1-4 Input grid level data Put all data on a monthly basis (only applies to wind and drought) Fill-in data holes (only applies to sea level data) Using the grid or station level data, compute monthly regional means for each components 17

54 ACI TM Calculation Steps 5 and 6 Using Monthly regional means compute seasonal means Average 3 consecutive months for seasonal means: Winter: Dec, Jan, Feb Spring: Mar, Apr, May Summer: Jun, Jul, Aug Fall: Sept, Oct, Nov For each component compute regional means and standard deviations for each month/season in the base period. 18

55 ACI TM Calculation Steps 7 and 8 Calculate component monthly/seasonal standardized anomalies X = (X X ref ) / σ ref (X) = ΔX / σ ref (X) Compute ACI Unweighted average of standardized anomalies ACI = (T90 T10 + P + D + W + S ) / 6 19

56 ACI TM Calculation Example Sea Level Example: REGION SOUTHERN PLAINS INPUT DATA Sept Oct Nov Station Galveston Port Isabel Rockport REGION MONTHLY MEANS REGION SEASONAL 4th QTR 2015 MEAN 7410 Sept Oct Nov REGION BASE PERIOD MONTHLY MEANS REGION BASE PERIOD MONTHLY STDEV REGION BASE PERIOD FOURTH QTR MEAN 7149 REGION BASE PERIOD FOURTH QTR STDEV 69 Standardized Anamolies Sept 2015 ( )/68 = 2.68 Standardized Anamolies Oct 2015 ( )/78 = 3.79 Standardized Anamolies Nov 2015 ( )/92 = 3.29 Standardized Anamolies 4th QTR 2015 ( )/69 =

57 ACI TM Results are In 21

58 Standardized T90, US and Canada 22

59 Standardized T10, US and Canada 23

60 Consecutive Dry Days, US and Canada 24

61 Extreme Precipitation, US and Canada 25

62 Wind Speed, US and Canada 26

63 Sea Level, US and Canada 27

64 Composite ACI Index, US and CAN. Unweighted average of standardized anomalies ACI = (T90 T10 + P + D + W + S ) / 6 28

65 Potential Uses Inform the public debate Educate actuaries and insurers Relevant to insured risk (property, liability, life and health) Incorporate climate trends in pricing Reflect higher risk into risk management Reconsider coverage and availability 29