Presentation at the BC Public Sector Pension Conference - Panel on Discount Rates and Disclosure of Public Sector Liabilities: Setting Assumptions for Funding Actuarial Valuations.
Jean-Claude Ménard, Chief Actuary, Office of the Chief Actuary, OSFI Victoria, 25 April 2013
Good afternoon. Thank you for inviting me here today to talk about a very complex topic of setting assumptions for funding valuations of pension plans. In my office - the Office of the Chief Actuary – this topic is addressed and discussed on a regular basis. In fact, setting assumptions is a part of our statutory obligations.
Office of the Chief Actuary (Slide2)
The Office of the Chief Actuary (OCA) consists of two sections. The first one is Social Insurance Programs Section, covering the Canada Pension Plan (CPP), the Old Age Security (OAS) program, Employment Insurance (EI) program, and the Canada Student Loans Program (CSLP). The second one is Public Sector Insurance and Pension Programs Section, that deals with federal public sector employee pension and benefits plans.
Our mandate is to conduct statutory actuarial valuations for these plans and programs. These valuations estimate the financial status of these plans and future funding requirements. All statutory actuarial reports that we prepare are tabled before Parliament by appropriate ministers.
As you can see, we are responsible for plans and programs that have impacts on almost every single Canadian.
Office of the Chief Actuary (cont’d) (Slide 3)
Actuarial staff of the OCA follows the Canadian Institute of Actuaries professional standards. Our work is regularly audited and reviewed both internally and externally to ensure its high quality. One of the most important external reviews is the peer review of CPP valuations. The independent panel of Canadian actuaries examines assumptions as well as underlying methodologies developed for the CPP valuations. It should be noted that these assumptions often form the basis for work done on other plans and programs falling under our responsibilities. The most notable examples are assumptions on future mortality improvements and real rates of return on asset classes.
The latest CPP peer review panel confirmed that CPP assumptions, both individually and in the aggregate, are within a reasonable range. All CPP external peer review reports are made public on OSFI’s Web site.
Funding valuation of the DB plan sets future contributions requirements (Slide 4)
Funding valuation of a DB plan determines how much plan’s members and sponsor should contribute in order for the plan to deliver promised benefits. The cost of the plan depends on future cash flows. For a typical DB pension plan, inflows consist mainly from contributions and investment returns. As such, future expected rate of return on assets is of the utmost importance. Outflows depend heavily on retirement benefits, thus it is important to reflect properly potential increases in life expectancies and the length of benefits payment period.
The Canadian Institute of Actuaries standards with respect to the funding valuations require the actuary to assume that the plan continues indefinitely, and to develop assumptions representing his or her best judgement as to future events (with provisions for adverse deviations, if necessary). This conceptual approach aims to stabilize the cost of the plan over time and to avoid dramatic changes in contributions requirements.
Main valuation assumptions (Slide 5)
Two main categories of assumptions used in the actuarial reports for pension plans are economic and demographic assumptions.
Main economic assumptions are inflation, real wage increases and real rate of return on assets. The ultimate OCA assumptions are shown on this slide. Demographic assumptions include fertility and migration rates (for the CPP and OAS only), mortality rates, salary increases, retirement, termination and disability rates etc. Some assumptions are developed using mainly plan experience (e.g. retirement rates), some take into account the overall economic and demographic environments (e.g. inflation). Finally, several assumptions are based on both plan’s specifics and general trends (e.g. real rate of return on assets and mortality).
Contribution to increase in life expectancy at birth has gradually shifted to people over age 65 (Slide 6)
Future mortality rates, especially at older ages, are a very important assumption for defined benefit pension plans. While prior to 1965, the rapid increase in life expectancy at birth was mainly due to a reduction in mortality rates for those under age 45, most of the increase in life expectancy now comes from reduction of mortality rates at ages 65 and older. Over the 20 years ending in 2005, more than half of the increase in life expectancy has been caused by a reduction in mortality rates after age 65. This trend is expected to become even more pronounced in the future.
Elderly mortality rates have decreased over the last 80 years, more so over the last 40 years (Slide 7)
For the age group 75 to 84, mortality rates have continually decreased over the last 80 years. The reduction was about 45% in the last 40 years (from 80 to 45 deaths per 1,000) compared to only 25% over the previous period of 40 years (from 107 to 80 deaths per 1,000). A further reduction of 35% is projected (from 45 to 29 deaths per 1,000). Moreover, mortality rates for Canadian elders are projected to fall much below those of their American counterparts. This is a remarkable achievement for the Canadian society but there is a price tag attached to it.
The OCA continuously strengthen mortality improvement rates assumption (Slide 8)
This chart illustrates the evolution of our projections of life expectancy of Canadian males aged 65. We constantly observe stronger mortality improvement rates than assumed. As a result, actuarial reports have shown subsequently lower projected mortality rates, and thus higher projected life expectancies. Overall, between the 2000 and 2009 reports, the projected age at death of a male aged 65 in 2025 has increased by almost 3 years from 83.6 to 86.3 years.
Setting Real Rate of Return assumption is one of the biggest challenges in the current environment (Slide 9)
One of the key assumptions used in actuarial valuations is the expected real rate of return on assets. This rate is used to discount future projected benefits in order to determine actuarial liabilities and current service cost.
We start by making an assumption with respect to the yield on long-term government bonds. The equity risk premium assumption is then used to derive an expected rate of returns on equities. Once real rates of return by assets class are developed, the assumed asset mix is used to obtain the real rate of return on a diversified portfolio of assets.
Government of Canada long-term marketable bonds. What did we know at the end of 2009? (Slide 10)
At the end of 2009, we were in the process of developing the real rate of return assumption for the 25th CPP actuarial report. At that time, the real yield on the Canada long-term marketable bonds was close to historical (dotted line). The average of private sector forecasts prepared for the federal government in December 2009 (red line) indicated that after a short period of lower values, the yield was expected to increase to 3.6% by 2015.
Given additional information available in early 2010 and applying our judgement, we made the assumption that the real yield on the Canada long-term bonds would be 2.1% in 2010 and would increase to its ultimate value of 2.8% by 2015.
So what has happened between 2009 and 2012? As you are aware, the yield decreased to an unprecedentedly low level.
Government of Canada long-term marketable bonds yields were at all-time low in 2011 (Slide 11)
The information presented on this slide is, generally speaking, a starting point to set the assumptions for the next CPP Actuarial Report as at 31 December 2012 that should be tabled before Parliament by the end of this calendar year.
In 2011 the real yield on Government of Canada long-term marketable bonds fell to 0.3%. Still, we continue to think that these yields will increase over the next 8 to 10 years. As you can see, recent projections produced by the Policy and Economic Analysis Program, the Rotman School of Management, University of Toronto, and by the private sector economists for the purpose of the 2013 Federal Budget are in line with our beliefs.
The future ERP is expected to be lower than in the past (3.4% for Canada) (Slide 12)
The equity risk premium assumption is used to derive an expected rate of returns on equities. For the past 113-year period, Canada and the world experienced an equity risk premium of 3.4% and 3.2%, respectively, whereas for the 50-year period 1964 to 2013, the ERP was only 0.9% for the world, and 1.0% for Canada. According to the Credit Suisse Yearbook 2013, historical equity risk premiums were higher than expected due to several non-repeatable factors (mainly diversification and globalization). As a result the long-term expected equity risk premium is likely to be lower than historical. Our long-term ERP assumption is 2.0% for Canadian and foreign developed market equities. An additional 1.0% is added for emerging markets.
Assumed real rate of return assumed by the OCA is in line with assumptions of peers (Slide 13)
The assumed real rate of return depends not only on assumptions on expected return for each asset class, but also on the asset mix. The equities usually have a higher assumed rate of return as well as a higher volatility. It should be noted that the assumed real rate of return for funding purposes is not necessarily the same as the target rate of return set by pension funds managers. Big pension funds increasingly use complex investment strategies to achieve or to exceed funding rate of return.
The current low interest rates environment results in an assumed average real rate of return of 3.6% for the next few years. The ultimate real rate of return assumed in the 25th CPP actuarial report is 4.0% per year.
Conclusion (Slide 14)
To conclude, I would like to highlight that in preparing actuarial reports for funding purposes, the OCA uses its best estimates to develop economic and demographic assumptions. There is an inherent uncertainty in estimating the cost of pension plans - actuaries do not have a crystal ball to see the future. As such, it is generally understood that experience gains and losses, as well as changes in economic and demographic environments, could result in plan costs that are different from projected. At the same time, in our opinion, this uncertainty should not be replaced by a certainty that interest rates will remain at their current low point over the long term.
Retirement is expensive and could become even more expensive in the future with improved longevity and uncertain future global economic growth.
Thank you for your attention and I am looking forward to a fruitful discussion.