1. Investment Regulations and DC Pensions Pablo Antolin, Financial Affair Division, OECD Asset allocation in uncertain time – CAMR Cass Business School, London, 2 July 2012

2. 2 Purpose Assess the impact of different quantitative approaches to regulate investment risk on the retirement income stemming from DC pension plans (voluntary and mandatory). Quantitative investment regulation affect retirement income through its impact on the choice of investment policies available.

3. 3 Structure of the presentation 1. Regulating DC pension plans: Why? 2. Description of different approaches to regulating investment risk in DC pensions. 3. Description of the stochastic model used to assess the regulatory impact on retirement income. 4. Discussion of the main results.

4. 4 Policy variable The policy objective variable used to assess the impact on retirement income of different quantitative approaches to regulate investment risk is: the replacement rate. the amount of pension benefits relative to the last wage given a certain contribution rate and a certain contribution period

5. 5 I. Regulating I risk in DC pension plans Policy makers take different approaches to the regulation of investment risk. The approach taken depends on: –Type of pension system being regulated (DB vs. DC; voluntary vs. Mandatory) –The broader country context (level of development K markets, level financial literacy, etc.) –The regulatory approach (use of quantitative rules as oppose to “pure” prudent person approach)

6. 6 Why regulate investment risk in DC plans? DC plans may expose individuals to excessive risk, in particular investment risk Raising possibility that pension benefits below expectations. Recent crisis has highlighted the volatility of RR in DC plans. Loss of confidence. In many countries pensions from DC plans are the main source of retirement income.

7. 7 OECD countries with large DC plans Countries with voluntary DC: Canada, Ireland, UK and the US. Countries with automatic enrolment (with opt-out): New Zealand, UK (in 2012). Countries with mandatory DC: Australia, Chile, Denmark, Hungary, Iceland, Mexico, Poland, Slovak Rep.

8. 8 II. Approaches to regulating investment risk in DC plans Quantitative approaches to regulate I risk:  Quantitative portfolio restrictions (limits to equities): Latin American countries, Hungary, Slovak Rep., Estonia  Minimum investment returns: Chile (minimum guarantee based on the average real rate of return of all pension fund previous 36 months), Switzerland (absolute rate of return guarantee: 2.75%), Belgium and Germany (nominal return equal to zero)

9. 9 Quantitative approaches to regulate I risk  Quantitative Risk ceiling: –Maximum Value at Risk (VaR): maximum loss in a portfolio with a given probability or confidence interval (e.g. 5%) and over a given planning horizon (e.g. day)  Mexico –Expected shortfall: average of the worst losses (insurance companies) –Stress tests, building worst case scenarios

10. 10 Quantitative approaches to regulate I risk considered in the report 1. Quantitative portfolio restrictions: limits to equities 2. Minimum investment returns: average nominal 2% (inflation) and 3.785% (wage growth). 3. Maximum Value at Risk (VaR) 4. Maximum replacement rate expected shortfall

11. 11 III. Model used: main pension assumptions Calculation assets accumulated in DC plans for people contributing 10% wages during 40 and 20 yrs. Retirement set at age 65. Pension benefits assuming people buy a life annuity at retirement (recent OECD work loosen this assumption).

12. 12 Stochastic modelling portfolio returns 4 asset classes: cash, bonds, equity, property 11 portfolios: Mean variance efficient –Portfolio 1: 0% in equities (5% cash,85% bonds,10 property) –Portfolio 11: 100% in equities ReturnVolatility (std) Cash4.0%2.0% Bonds5.5%4.5% Equity7.5%15.0% Property6.0%10.0%

13. 13 Stochastic modelling portfolio returns 3 investment strategies: –fixed or balanced portfolio: initial allocation remains constant over time. –Dynamic risk budget: each portfolio changes asset allocation according to its risk budget. –Life cycle: last 10 years reduces equity exposure to reach 0% at retirement. 11 portfolios and 3 investment strategies result in 33 investment policies

14. 14 Stochastic modelling: Simulations Stochastic simulations of returns each asset class  stochastic simulation of portfolio returns Stochastic inflation and life expectancy (annuity rate). Given contributions and contribution period: 10,000 MC simulations of replacement rates for each of 33 IP

15. 15 Main outcome of the modelling 10,000 MC simulations of replacement rates for each of 33 IP Distribution function of RR: –mean, –median, –the standard deviation and –different percentiles, in particular the 5 th and the 95 th percentile.

16. 16 Assessment measures: Risk Risk criteria to assess different investment policies is the 5 th percentile: –Probability that replacement rates above that value equals 95% –Probability that RR are below is 5% Policy makers and regulators may prefer investment policies that provide a higher replacement rate with only a five percent probability that the replacement rate may fall below.

17. 17 RR by level of risk and investment policy (40 yrs)

18. 18 RR by level of risk and investment policy (40 yrs)

19. 19 RR at the 5 th percentile (40 yrs)

20. 20 Median RR (40 yrs)

21. 21 IV. Main results Trade-off between RR expectations and risk. Very low allocations to equities ( 80%) look unattractive in this context. In btw wide range of options for regulators and supervisors to consider Risk adverse regulators/individuals  conservative investment policies: no more than 30% risk assets and life cycle investment strategies (optimal?). However, large loss of potential RR as a result. Less risk aversion higher expected RR.

22. 22 Regulatory Approaches How different regulatory approaches affect the investment policies available? Limits on equities VaR: investment policies with a monthly portfolio return below -2% in more than 5% cases excluded. Minimum return: –2% nominal  RR=22% (40yrs); RR=13% (20). –Wage growth (3.8%)  RR=31% (40); RR=16% (20). RR expected shortfall: 95% RR below minimum RR (e.g. 25%) are within 5 pp of the minimum RR

23. 23 Quantitative I limits (max. 30% equities) Steer pensions funds and people to investment policies with a large share in bonds, reducing the downside risk, but also reduces potential gains in RR Setting quantitative limits could be efficient a priori (if you believe your model is right) but not a posteriori (you model may not be confirmed by real events)

24. 24 Value at Risk (-2% in 5% cases) This short-term investment return regulation steers pension funds and investors to investment policies with no more than 30% in risk assets (equities). Complicated way of achieving same results that limits on equities? The investment policies available under this regulatory framework are independent of the length of the contribution/accumulation period

25. 25 Min returns, Min RR, Expected Shortfall The availability of different investment policies under these regulatory approaches depends on the severity or strictness of the probability threshold. It also depends on the length of the contribution and accumulation period.

26. 26 Min returns, Min RR, Expected Shortfall Relatively high minimum returns (e.g. wage growth, 3.785%) or RR (e.g. 25%) render all investment policies unavailable when individuals only contribute around 20 yrs instead of 40 yrs. For lower minimums and longer contribution periods which investment policies are available depends on the probability threshold –99.5%  very conservative I policy (bonds>70%) –80% wider range of options and higher potential RR

27. 27 Minimum rate of return (2%) – 40yrs

28. 28 Minimum rate of return (2%) – 20yrs

29. 29 Minimum rate of return (3.785%) – 40yrs

30. 30 Minimum rate of return (3.785%) – 20yrs

31. 31 RR expected shortfall of 5 pp below a 25% minimum RR – 40 years

32. 32 RR expected shortfall of 5 pp below a 25% minimum RR – 20 yrs

33. 33 Conclusions Choosing among different investment policies requires balancing the trade-off between higher potential retirement income and risks associated Risk adverse regulators & supervisors will aim for policies reducing the downside risk or that minimise the risk of unfavourable outcomes from DC plans. This would require moving into relatively conservative investment policies, where the share of assets allocated in bonds is quite large (≥ 60%).

34. 34 Conclusions Important to stress that there is not a single correct trade-off, depends on the country context and the risk aversion levels. Countries where payments from DC pension plans are the main source of retirement income, the cost to the society of downside risks or unfavourable outcomes is much larger than in countries where they have other sources of retirement income (public pensions).

35. Thank you! www.oecd.org/daf/pensions www.oecd.org/daf/fin/wp pablo.antolin@oecd.org

36. 36 Replacement rates at the 5 th percentile

37. 37 Replacement rates at the 95 th percentile