Public Pension Finance Symposium May Session 5: Risk Management Ideas for Public Pensions. Moderator: Mark Ruloff

Transcription

1 Public Pension Finance Symposium May 2009 Session 5: Risk Management Ideas for Public Pensions Moderator: Mark Ruloff

2 watsonwyatt.com PUBLIC PENSION PLAN FINANCE SYMPOSIUM Risk Management Ideas for Public Pensions Mark Ruloff May 19, 2009 My Views Going In Stochastic ranges may be used to make risk seem acceptable 5% chance needs to be thought of as 1 in 20 year event, when not if Need to measure/accumulate probability of ruin over time Underfunding, because of liability mis-measurement, results in the need for more risk taking in investment strategy More aggressive mixes then appear to be lower risk (Saving $1/yr for $1M retirement, buy lottery tickets as improves chances) Aggressive investment can only attempt to compensate for underfunding Difference between term premium in bonds and risk premium in stocks Smoothing defers problem Smoothing only helps with market noise, not market trends as it delays needed action Can push ruin out past the end of forecast period so it only appears to improve things Article, Defined Benefit Plans are More Successful with Bonds 2

3 Communicating Risk in Public Plans - McCrory What I liked Full simulation or stochastic modeling Showing actual trial results, Note: Many trials cross cost of twice current but at different times Back testing Simple projections What I didn t like Quartiles and range projection Therefore, we can say with 50% confidence that the cost will range from nothing to twice the current levels This is just in any one year, examining trial results over entire forecast shows a different story, more than 50% are twice at some point during forecast Risk free rate not relevant Underfunding is root of risk Don t get hung up on asset models Many models do not reflect the benefits of cash flow matching bonds 3 Public Plans: Using Risk Profiles to Manage Funding Goals- Davis, Porter & Steffen What I liked Funding cushion But may be used for benefit increases instead Modified LDI Recognizes benefits of cashflow matching But ignores duration matching issues with MVL and true LDI What I didn t like Projected benefit payments are modeled as if they were bonds and then are assigned the market value of equivalent bonds. Because they are as secure as bonds Not like an equity dividend that can be cut Investment risk Investment risk is being taken to make up for lower funding which took advance credit for future, hoped for, equity risk premiums 4

4 How Much Investment Risk Can A Government Sponsored Pension Plan Afford? - Hallmark & Kelly What I liked Ranking, Feedback loop problem Different risk characteristic for different plan situations This approach further seems to be based on the assumption that the sponsor has an infinite appetite for risk as long as they expect to be compensated over the long-term. Plan event horizon, like probability of ruin measure What I didn t like Long term investors Term premium differs from risk premium No accumulation of plan event horizon over longer term (100yrs) Financial economics purists argue that public plan sponsors should have minimal appetite for investment risk, resulting in a very different one-size fits all investment policy. Overlooks pass through nature of public fund to taxpayers and Tax arbitrage as primary financial economic argument here 5 Projections How long? Where is the focus? $70 $60 $50 $40 $30 XYZ Company 1 Contributions ($000) - 70% Eq 30% Agg Bonds Contributions based on 5 year smoothing of asset gains and losses as well as 30 year amortization of the gains and losses that are not smoothed. $20 $10 $ Year 1st-5th percentile 5th-25th 25th-50th 50th-75th 75th-95th 95th-99th 50 th percentile results look fine 95 th percentile are high but only 5% chance of that However, that is 5% each year. What is chance of this happening sometime 50% perhaps? Not a question of if but when 6

5 Ruin: Chance Assets are Depleted XYZ Company 1 MV<BP - 70% Eq 30% Agg Bonds 50% 40% 30% Mean 20% 10% 0% Year Lower returns over longer time periods cause assets to be depleted. Although few trials are depleted in any one year, many trials are depleted over time Liability Measure Takes Advance Credit for Future Equity Risk Premium Vast amount of problems (98% as measured in my article, ), is cause by liability calculations anticipating equity risk premium rather than actually investing in equities Equities are expected to earn more because they involve more risk Downside risk and not just noise Liability includes equity risk premium in discounting But $100 of bonds is as valuable as $100 of stocks Expected return is just fair compensation for additional risk Market value of excess return is zero 8

6 Limitations and Disclaimers The opinions expressed in this presentation are not necessarily the position of my employer. While I hope that you find this presentation informative, please note that this presentation is provided for informational purposes only and may not be relied upon in any manner. 9