2. 1/17/2016 Strategic Asset Allocation 1 Demography, Capital Markets and Pension Risk Management session 2 Andrei Simonov

3. Analysing pension risk Illustrative example £ million LIABILITY EXPOSURES Unrewarded (no risk premium) ASSET RETURN RISKS Rewarded 1 in 20 chance that deficit is at least £Xm higher than expected Y% probability of being at least 100% funded in 10 years time Longevity hedging should be considered alongside other risk mitigation techniques using risk return framework

4. 3 Demographic factors at work Increasing longevity Lower fertility Retirement of Baby Boom

5. 4 From pyramids to columns 0 - 4 5 - 9 10 - 14 15 - 19 20 - 24 25 - 29 30 - 34 35 - 39 40 - 44 45 - 49 50 - 54 55 - 59 60 - 64 65 - 69 70 - 74 75 - 79 80 - 84 85 - 89 90 - 94 95 - 99 100 + Age Group B A A B A B

6. 5 People over SPA to those aged 20 – SPA* * SPA: State Pension Age With SPA fixed at 65 With SPA rising proportionally (to 68.5 in 2050 and 70.2 in 2070) 1 (1) This proportionate adjustment maintains the proportion of life over 20 years old which is spent in retirement at 27.5%

7. 6 Lower fertility – The inherent challenge to pension systems Increased ratio of pensioners to contributors Savers of generation 1 have to sell accumulated assets to “smaller” * generation 2 Transitional asset price fall effect K/L rises: return on capital falls Lower pensions relative to average earnings Increase Pension Age more than proportionally with life expectancy Higher contribution rates PAYG Funded * Smaller can mean either absolutely smaller than G1 (if fertility  2.0) or “smaller than would be the case if fertility had not fallen”

8. 7 Possible de facto demographic effects on funded systems and capital markets Transitional asset price fall effect (at sale) K/L rises: return on capital falls Transitional asset price rise effect Longer-term effect; K/L rises, return on capital falls Not inherent but could occur if future pensioners do not adjust retirement ages but instead increase savings rate Inherent effect of shift to lower fertility Lower Fertility Increased Longevity

9. 8 Demographic impacts on returns to capital Garry Young:Baby-boom generation -0.1% Increased longevity-0.1% Falling fertility-0.3% David Miles:Given future actual trends in UK demographics, returns fall:  4.56% (1990) to 4.22% (2030)  4.56% (1990) to 3.97% (2060) if PAYG phased out Model Results

10. 9 Real S&P500 price index and % of 40-64 year olds among total U.S population 1950-2003 Source:: Poterba (2004), with additional data to 2006

11. 10 Theoretical & empirical approaches to measuring demographic effects “Given the limited amount of time series on returns and demographic variation, and the difficulty of controlling for all of the other factors that may affect asset values and asset returns, the theoretical models should be accorded substantial weight in evaluating the potential impact of demographic shifts” Poterba: “The Impact of Population Ageing on Financial Markets”

12. 11 Global glut of savings hypothesis Fewer children enable higher savings rate Awareness of greater longevity, fewer children and lack of social welfare net, require a high savings rate Global glut of savings relative to investment Long-term, not just cyclical, fall in real interest rates In China and other East Asian countries Developed countries save more to cope with their demographic/pension challenges Transitional positive asset price effects

13. 12 Real yields to maturity on UK index-linked gilts 1986 – 2004

14. 13 UK Long-term real interest rates Source: Morgan Stanley Research

15. 14 Whole world gross savings rate 1981 - 2005 Source: IMF World Economic Outlook database

16. 15 Gross savings rates: developing Asia and the US % of GDP 1981 - 2005 Source: IMF World Economic Outlook database USA Developing Asia

17. 16 Demographic challenges to funded pension systems Increasing longevity Lower fertility Uncertainty of longevity forecasts Not inherent problem but possible de facto Overwhelmed in short-term by globalisation effect

18. 17 Survivor Products: Managing longevity risk & mortality improvements

19. 18 Current Forces Affecting the Size and Ownership of Longevity Risk The Mosaic Today Retirement population growing Longevity risk moving from corporates to individuals Cost of longevity significant and rising Inflation could exacerbate longevity costs Longevity extending Credit Crunch moving longevity risk arguably to Government The UK population of retirees (i.e. people 65+) is set to increase by 60% by 2032 from 10 million to 16 million due mainly to the ageing of the baby boom generation Longevity continues to increase for retirees at historically high rates against largely fixed retirement / entitlement dates The current cost of life extension in the UK is estimated at £12.5 to £24.7 billion per year Current fiscal and monetary policy may be sowing the seeds of high inflation and expectations of inflation Strong forces are causing corporates to close existing Defined Benefit (DB) schemes and transition to Defined Contribution / Personal Account schemes The Credit Crunch is causing a significant increase in DB pensioner risk to move to the Pension Protection Fund (PPF)

20. Source:Pension Corporation LLP.. 19 Moving Longevity Risk: Recent Transactions (Europe) DateRisk Owner Size (GBP millions) Type of Longevity Risk Risk Recipient January 2008LucidaNot disclosedLife annuitants Capital markets investors July 2008Canada Life£500Life annuitants Capital markets investors February 2009Abbey Life£1,500Life annuitantsReinsurers March 2009Aviva£475Life annuitants Reinsurers and capital markets investors June 2009 Babcock International £500 – £750PensionersReinsurers July 2009RSA£1,900Pensioners Pension insurer and reinsurers September 2009 (announcement) Royal County of Berkshire [£500 – £750]Pensioners[Pension insurer] Total (Jan 08 – Sep 09) £5,375 – £5,875

21. 20 Nothing is certain in life except death and taxes (B Franklin). Over last 20 years, it has become clear that, while death is no less inevitable than before: –it is getting later –and its timing has become increasingly uncertain. The problem

22. 21 When British welfare state began in 1948, men could draw their state pension at 65 and expect to live until 67 and only a few lived beyond 70. At beginning of 21st Century, British men can still draw their pension from age 65 but now live into their early 80s. Significant proportion of women living into their late 80s. The problem

23. 22 Mortality improvements over time

24. 23 What is longevity risk? (Broken limits to life expectancy – Oeppen & Vaupel)

25. …while longevity costs increase Source: AEGON

26. 25 Evident for many years that mortality rates have been evolving in apparently stochastic fashion. Sequences do exhibit general trend, but changes have an unpredictable element: –not only from one period to next –but also over the long run. Stochastic nature of mortality improvements

27. 26 Large number of products in life insurance and pensions have mortality as key source of risk. Products exposed to unanticipated changes over time in mortality rates of relevant reference populations. Eg annuity providers exposed to risk that mortality rates of pensioners will fall at faster rate than accounted for in pricing and reserving calculations: –Current pool of annuitants living 2 years longer than anticipated Longevity risk

28. 27 Annuities are commoditised products selling on basis of price, profit margins have to be kept low in order to gain market share. If mortality assumption built into price of annuities turn out to be gross overestimate, cuts straight into profit margins of annuity providers. Most life companies claim to lose money on annuity business. Longevity risk

29. 28 Yet life annuities are mainstay of pension plans throughout the world: –they are the only instrument ever devised capable of hedging longevity risk. Without them, pension plans will be unable to perform their fundamental task of protecting retirees from outliving their resources for however long they live. Real danger that they might disappear from financial scene. Longevity risk

30. 29 Equitable Life: Embedded options in annuity contracts became very valuable in 1990's due to combination of falling interest rates and improvements in mortality. Problems avoided if EL could hedge exposures to: –interest-rate risk –mortality improvement risk. Longevity risk

31. 30 Longevity risk in UK pension provision, £billion of total liabilities- broad estimates: end 2003 Figure 5.17 p181

32. 31 Reinsurers (eg Swiss Re) have stopped reinsuring longevity risk of life offices! Significant concern!

33. 32 Long-dated survivor bonds: Life annuity bond: coupon payments decline in line with mortality index: –Eg based on population of 65-year olds on issue date. As population cohort dies out, coupon payments decline, but continue in payment until the entire cohort dies. Eg, if after one year 1.5% of population has died out, 2 nd year’s coupon payment is 98.5% of 1 st year’s etc Survivor Products

34. 33 Bond holder, eg life office writing annuities, protected from aggregate mortality risk it faces. Based on Tontine Bonds issued by European governments in 17th and 18th centuries Recently revived by Blake and Burrows (2001) and Lin and Cox (2004). Survivor Products

35. 34 November 2004 Issuer: European Investment Bank (AAA) Issue: £540m, 25 year Mortality index: 65 year-old males from England & Wales (ONS) Structurer/manager: BNP Paribas (assumes longevity risk) Reinsurer of longevity risk: PartnerRe, Bermuda Investors: UK pension funds BNP Paribas Longevity Bond

36. 35 BNP Paribas Longevity Bond

37. 36

38. 37 Provides better match for liabilities of pension funds and life insurers than other available investments: – other than purchasing (re)insurance to cover the longevity risk (i.e annuities) Bond also provides long term interest rate hedge. Longevity index transparent EIB has AAA credit rating. Life insurers holding longevity bond as hedge may be able to hold lower prudential margins. Advantages of longevity bond

39. 38 Longevity BondAnnuity Partial hedging of the longevity risk Full hedging of longevity risk Low credit risk of EIB (rated AAA) Higher credit risk of the insurer but there is additional protection through the government compensation scheme Fixed term of 25 years Covers the full term of the liability Only level pensions matched Different annuities can be used to match non - level pensions

40. 39 Short-dated, mortality-linked securities: Market-traded securities whose payments are linked to mortality index Similar to catastrophe bonds (Schmock, 1999, Lane, 2000, Wang, 2002, and Muermann, 2004) Survivor Products

41. 40 Designed to securitise Swiss Re’s own holding of mortality risk! 3-year contract (matures 1 Jan 2007) which allows issuer to reduce exposure to catastrophic mortality events: –severe outbreak of influenza –major terrorist attack (WMD) –natural catastrophe. Mortality index (MI): –US (70%), UK (15%), France (7.5%), Italy (5%), Switzerland (2.5%). –Male (65%), Female (35%) –Also age bands Swiss Re Bond 2003

42. …driving the need for de-risking… Source: CBS data, AEGON

43. 42 All resulted from avian flu virus mutating with human flu virus 11 outbreaks in 300 years 1580 –First confirmed flu pandemic 1782 –Summer Flu –Started in China –Hit young adults 1889 –Russian Flu –Over 20% of world population infected –1m deaths Influenza pandemics

44. 43 1918-19 –Spanish Flu –Started in Kansas –Killed 50m people worldwide: 250,000 in UK –More than died in WW1, in shorter period –20% of world’s population infected and 1% killed –Spread along trade routes and shipping lines Influenza pandemics

45. 44 1957-58 –Asian Flu –2m deaths –Hit teenagers hardest – Spread around world in 6 months 1968-69 –Hong Kong Flu –Started in China –1m deaths –Spread slowly with moderate symptons Influenza pandemics

46. 45 2005-06 –Started in SE Asia –H5N1 virus –Closely related to 1918 Spanish virus Influenza pandemics

47. 46 $400m, principal at risk ‘if, during any single calendar year, combined mortality index exceeds 130% of baseline 2002 level’. Principal exhausted if index exceeds 150% Equivalent to a call option spread on the index with: –Lower strike price of 130% –Upper strike price of 150% Investors get quarterly coupons of 3-mo USD Libor + 135bp Swiss Re Bond 2003

48. 47 Swiss Re Bond 2003

49. 48 Swiss Re Bond 2003

50. 49 Bond valued using Extreme Value Theory (Beelders & Colarossi (2004)) Assume Generalised Pareto Distribution Probability of attachment: –P[MI(t)>1.3MI(2002)] = 0.31% Probability of exhaustion: –P[MI(t)>1.5MI(2002)] = 0.15% Expected loss = 22bp < 135bp A good deal for investors! Bond trading at Libor + 100bp in June 2004 Swiss Re Bond 2003

51. 50 Survivor swaps: Counterparties swap fixed series of payments in return for series of payments linked to number of survivors in given cohort: –UK annuity provider could swap cash flows based on UK mortality index for cash flows based on US mortality index from a US annuity provider counterparty –Would enable both counterparties to diversify their longevity risks internationally. Dowd et al (2004) Survivor Products

52. Q-forward (index swap) Single contract on mortality index observation at a given point in time –For example, observed mortality rate on 61-70 year old males in 10 years time –Expected rate is – if we extrapolate – 1.0%. This is an improvement from the current rate which is roughly 1.4%. –In addition, a risk premium will be collected, so the fixed rate may be 0.9% (for illustration only – not indicative) –The payout will be a notional amount times the differential between the actual rate 10 years from now and the fixed rate A challenge with q-forwards is ensuring that enough are purchased to reasonably describe the portfolio and avoiding basis risk between the plan and the index

53. Indemnity swap The key “complicating” factor for an indemnity swap is that it requires the risk assuming entity to value and underwrite each risk it is assuming individually. Clearly, it is more efficient (and therefore possibly cheaper) for the assuming entity to analyze an index and then trade on that many times over However, an indemnity swap does not have to be (much) more complicated than an index swap –Cover specified benefits (including survivors) as life annuities. Then we only need to know who is alive at any point in time. –An exit clause – possibly with an exit fee – will add liquidity to allow the non-exiting party re-establish the position with another counterparty –Valuation will be marked-to-model as is also the case with an index swap

54. 53 Annuity futures: Prices linked to specified future market annuity rate Mortality options: Payout depends on underlying mortality table at payment date. Eg, EL guaranteed annuity contract Survivor Products

55. 54 Reference population underlying calculation of mortality rates central to both: –Viability –Liquidity of contracts. Hedging demand from investors (eg life offices) wishing to hedge mortality exposures. If reference population v different from investor’s specific population, then investor will be exposed to significant basis risk: –Might conclude that mortality derivative is not worth holding. Demand side of market

56. 55 Speculative demand: –depends on liquidity. Adequate liquidity will require small number of reference populations: –Need to be chosen carefully to ensure that level of basis risk is small for investors with hedging demands. Demand from hedge funds: –seeking instruments that have low correlation with existing financial instruments Demand side of market

57. 56 Government: –Securitising social security budget Corporates long longevity risk: –Pharamceuticals Supply side of market

58. 57 After more than year, BNP Paribas longevity bond had not generated sufficient demand to be launched: –Has been withdrawn for redesign Suggests significant barriers need to be overcome before sustainable market in survivor products and derivatives emerges. Barriers to development in cash market

59. 58 Reasons why BNP bond did not launch: –design issues which make bond an imperfect hedge for longevity risk –pricing issues –institutional issues Barriers to development in cash market

60. 59 Small scheme will find it difficult to use bond to match its liabilities: –as variance between actual and expected mortality will be quite large. Mortality experience of individual pension funds and life insurers may be different from reference UK population. Bond only provides hedge for longevity of males: –pension funds and life insurers also exposed to significant longevity risk from females. Design issues

61. 60 Liabilities for pension funds and life insurers give greater weight to the lives receiving larger pensions. Further, significant differences in mortality of those receiving larger pensions compared to those receiving lower pensions. As payments under bond effectively give equal weight to all the lives in the UK population, the already imperfect hedge provided by the longevity bond is worsened. Design issues

62. 61 Bond only matches cashflow under level pension –while large portion of pensions paid by pension funds and life insurers will be increasing at RPI/LPI/CPI. Bond is progressively worse hedge for pension liabilities related to younger or older cohorts. Design issues

63. 62 Need to forecast mortality index MI’s Need to estimate r’s Correlation between MI and r: –Anticipated to be low Pricing issues

64. 63 Above model valid only in complete market In incomplete market, need to convert projected deterministic mortality rates into risk-neutral probabilities –E.g using Wang transform –Lin & Cox (2005) Pricing issues

65. 64 Longevity risk premium built into initial price of bond set at 20 basis points. Given that this is first ever bond brought to market, markets have no real feeling as to how fair this figure is. However, concern that up-front capital was too large compared with risks being hedged by bond: –longevity and interest rate risks leaving no capital for other risks to be hedged –e.g. inflation Pricing issues

66. 65 Issue size too small to create liquid market. Consultants reluctant to recommend it to trustees. Fund managers do not currently have mandate to manage longevity risk. Fund managers have not welcomed bond: –since believe it would be closely held and they would not make money from it being traded. Partner Re is unlikely to be perceived as being a natural holder of UK longevity risk. Institutional issues

67. 66 Last point highly significant Reflects view that key determinant of future issue of longevity bonds is availability of sufficient reinsurance capacity. Neither UK-based nor EU-based reinsurer willing to provide cover for BNP bond Partner Re not prepared to offer cover above issue size of £540m. Has been questioned whether EU’s solvency requirements render reinsurance cover within EU prohibitively expensive. Institutional issues

68. 67 Following factors key to success of particular futures contract: –defined as having consistently high volume of trade and open interest: Must be large, active and liquid spot market for underlying with good price transparency: –by far the most important factor: –indeed no futures contract has ever survived without a spot market satisfying these conditions. Barriers to development in futures market

69. 68 Spot prices must be sufficiently volatile to create hedging needs and speculative interest. Relative hedging demand can be measured by level of open interest relative to volume: –since former excludes the many speculators who do not hold overnight positions. Low open interest to volume ratio is an indication of high liquidity : –another sign of successful futures contract. Barriers to development in futures market

70. 69 Underlying must be homogeneous and/or have well-defined grading system. Market in underlying must not be heavily concentrated on either buy or sell side: –since this can lead to price manipulation. Futures contract must be effective in reducing risk. Barriers to development in futures market

71. 70 Liquidity costs: –bid-ask spreads –execution risk: risk of adverse price movements before trade execution Liquidity costs in the futures contract must not be significantly higher than those operating in any existing cross-hedge futures contract. Barriers to development in futures market

72. 71 CPI futures contract listed on US Coffee, Sugar and Cocoa Exchange in June 1985 Delisted in April 1987 with only 10,000 contracts traded. Reasons for failure: –no inflation-linked securities market at the time. –underlying was infrequently published index. –no stable pricing relationship with other instruments. Failure of futures contracts

73. 72 Futures contract on Treasury inflation- protected securities (TIPS) listed on Chicago Board of Trade in June 1997 Delisted before end of the year with only 22 contracts traded. Reasons for failure: –TIPS had only started trading five months before. –Only a single 10-year TIPS outstanding. –Futures contract competed with underlying for liquidity. –Uncertainty over fate of TIPS programme. Failure of futures contracts

74. 73 CME launched CPI futures contract in February 2004 which is still trading. Reasons for survival: –Inflation-linked securities have gained acceptance amongst investors –TIPS have evolved into recognised asset class. –Well understood pricing relationships allowing for arbitrage possibilities between TIPS, fixed-interest Treasury bonds and CPI futures. –US Treasury committed to long-term TIPS issuance. –CPI futures do not compete directly with but rather complement TIPS and uses same inflation index. –Contract traded on Globex electronic trading platform: automated two-sided price quotes from lead market maker. Failure of futures contracts

75. 74 Sufficiently large, active and liquid spot market in longevity bonds must be established well before any futures market started. Mortality index behind longevity bond must be fair estimate of true mortality and have minimal time basis risk: –CPI index suffers from same potential problems –so survival of CPI futures contract on CME suggests these problems can be overcome. Important lessons for development of mortality-linked futures market

76. 75 Although mortality indices are calculated infrequently, spot prices of longevity bonds likely to exhibit high degree of volatility on account of bonds’ high duration. Underlying mortality indices must be few in number and well-defined: –small number of contracts helps to increase liquidity –but also leads to contemporaneous basis risk arising from different mortality experience of population cohort covered by mortality index and cohort relevant to hedger. Important lessons for development of mortality-linked futures market

77. 76 Potential weak point in longevity bond market is on the supply side: –since few natural issuers on supply side. Futures contract would be effective in reducing aggregate risk,: –but small number of mortality indices might well leave substantial basis risk. No reason to suppose liquidity costs in futures contract would be any higher than for other bond futures contracts. Important lessons for development of mortality-linked futures market

78. 77 Existence of survivor products: – will facilitate the development of annuities markets in the developing world – and could well save annuities markets in the developed world from extinction. Essential to prevent annuity providers going bust! Conclusion

79. 78 If survivor products fail to be issued in sufficient size: – either the state (i.e., the next generation) is forced to bail out pensioners –or companies withdraw from pension provision –or insurance companies stop selling annuities – or pensioners risk living in extreme poverty in old age, having spent their accumulated assets Conclusion