Reaching for yield or resiliency? Explaining the shift in Canadian pension plan portfolios

Introduction

“Desperate for returns, pension funds have piled into private markets in recent years” (Taking Back Control 2020). According to this claim—formally known as a reach for yield—pension plans meet the need for adequate returns by looking beyond the traditional portfolio mix of 60 percent stocks and 40 percent bonds.

We study the reach-for-yield claim in the context of Canadian private pension plans between 1998 and 2018. The Office of the Superintendent of Financial Institutions (OSFI) provided anonymized data on the assets and liabilities of 128 federally regulated, private, defined benefit pension plans.

Our analysis reveals that, as the level of interest rates declined, 64 percent of plans in our sample shifted away from the traditional 60/40 mix. In its place, they opted for portfolios with more alternative assets, such as private equity, real estate and infrastructure, but a majority allocation in bonds. For one-third of these plans, the portfolio overhaul is substantial—representing more than 30 percent of all assets.

The larger allocation to bonds—with safer and lower returns—is hard to square with reaching for yield. Instead, we argue that managing solvency risk can explain this broad shift.

For a long time, regulatory solvency requirements have applied to many corporate pension plans around the world—including those in Canada that are regulated by OSFI (Greenwood and Vissing-Jorgensen 2019). When plans fall short of their solvency requirements, sponsoring employers must make special cash payments beyond their regular annual contributions to the plans. This becomes a costly risk. We refer to this risk as solvency risk.

We show that the secular decline in interest rates since 2004 has eroded solvency surpluses. In response, the shift toward bonds:

• protects pension plans against exposure to interest rates
• reduces solvency risk

At the same time, to partly offset the lower returns of government bonds, these plans shifted their portfolio to corporate bonds, private equity, private infrastructure and real estate. The total portfolio shift has resulted in a small decrease in expected returns, by about 25 basis points, compared with plans that maintained the more traditional 60/40 split.

Despite the decline in interest rates, most of the plans that overhauled more than 30 percent of their portfolio returned to 100 percent funded status by 2016, and their solvency ratio volatility declined by 30 percent. In contrast, most plans that kept the 60/40 portfolio mix ran a solvency deficit in 2018.

Our findings provide a new and complementary explanation for pension plans’ inclusion of more alternative assets in their portfolios. In the standard reach-for-yield explanation, plans respond to lower yields by increasing their overall risk exposure so they can meet their return targets.

We turn this explanation on its head. It is true that, in isolation, investing in alternative assets and corporate bonds or using financial leverage means taking specific risks in exchange for higher returns. But we show that these investments are part of a broader strategy that aims to reduce overall solvency risk.

We find that the likelihood of a portfolio shift increases with plan size, presumably because large plans have greater ability to invest in alternative asset classes than small plans do. In our sample:

• 85 percent of plans whose assets under management (AUM) exceeded $0.5 billion shifted
• 52 percent of plans with AUM below $0.5 billion shifted

Plan size is not the only factor, though. We find that exposure to solvency risk is another key predictor of shifting portfolios.

Plans that do not shift may face more solvency risk. These plans may have good reasons for maintaining the 60/40 portfolio mix but doing so leaves them exposed to short-term fluctuations in asset values and interest rates. Keeping the traditional mix therefore means that plan sponsors must stand ready to make special payments to the plans if solvency deficits persist.

Sustained decreases in interest rates have eroded plan solvency surpluses

We use anonymized data provided by OSFI and focus on the 128 pension plans that:

• were operational in 2018
• managed over $50 million in assets (details in the Appendix)

These plans represent a diverse group that together manage $200 billion in assets, which is equivalent to Canada’s third largest pension plan (Bédard-Pagé et al. 2016).

OSFI requires pension plans to include two assets-to-liability ratios—the solvency ratio and the going-concern ratio—that must have a value greater than one.

For both ratios, the plan’s asset value is the numerator, and the current value of future pension liabilities is the denominator. One key difference between the two ratios is the method used to discount liabilities.

The solvency ratio uses a long-term bond yield to discount liabilities. This ensures that plans have sufficient assets to purchase an annuity that covers the promised pension benefits if the employer terminates the plan in the near future. Because the solvency discount rate is a market rate, the solvency requirement is highly sensitive to fluctuations in interest rates.

In contrast, the going-concern ratio uses the expected rate of return of the plan’s assets as to discount liabilities, so it is much less sensitive to fluctuations in interest rates. This ratio ensures that a plan’s assets can cover its liabilities if the plan remains active over the long term.

Chart 1 shows that pension plans ran solvency and going-concern surpluses in the late 1990s (illustrated by the area above the horizontal line). It also shows that the solvency discount rate gradually declined from approximately 6 percent in 2004 to almost 3 percent in 2018, consistent with the continuous decline of interest rates during this period (Feunou and Fontaine 2019). The steady decline in interest rates starting in 2004 has:

• raised the valuations of solvency liabilities
• eroded solvency surpluses

By itself, the pattern of the discount rates explains 42 percent of the declines in the median solvency ratio. In fact, many Canadian plans have experienced a persistent solvency deficit since 2002 because of the interest rate decline.

The lower interest rates have less of an impact on going-concern ratios. These ratios have therefore returned to surpluses since 2013, in response to the strong recovery of plan assets after the 2008–09 global financial crisis.

Chart 1: Pension plan solvency surpluses turned into deficits as interest rates declined

Note: The solvency and going-concern ratios (left scale) and the solvency discount rate (right scale) are the median of all 128 plans.
Source: Office of the Superintendent of Financial Institutions pension fund dataLast observation: 2018

Pension plans respond to low interest rates by shifting their portfolios

To examine the portfolio response of Canadian private pension plans to the decline in interest rates, we split their assets into the four largest classes:

• stocks
• corporate bonds
• real estate
• other, which includes mostly private equity and infrastructure

We then calculate the change in each asset class allocation between 2004 and 2018 and sum these changes to obtain the total portfolio shift for each plan (details in the Appendix). We select 2004 as the starting year because it coincides with the beginning of the sustained low interest rate environment. A portfolio shift of 0 percent indicates no changes to the asset allocation, whereas a shift of 100 percent indicates a complete overhaul of the portfolio.

Chart 2 shows that 64 percent of pension plans in our sample shifted their portfolio allocation by more than 10 percent since 2004. These plans accounted for 88 percent of total plan assets in 2018. For about one-third of these plans, the shift is substantial, representing more than 30 percent of the portfolio. We refer to these plans as large shifter plans and the other plans as moderate shifter and non-shifter plans.

Chart 2: Pension plans shifted their portfolio allocations between 2004 and 2018

Source: Office of the Superintendent of Financial Institutions pension fund dataLast observation: 2018

Chart 3 plots the median asset allocations of each group over time. We see that:

• all plans began with the same traditional 60/40 allocation in 2004
• both moderate and large shifter plans gradually transformed their portfolios as interest rates persistently declined, indicating a structural shift
• both moderate and large shifter plans moved their portfolios in the same direction, decreasing stock allocation and increasing allocation of alternative assets and bonds
• while large shifter plans changed their portfolios starting in 2004, the typical moderate shifter plan started in the 2010s

Chart 3: Plans that shifted since 2004 now hold a majority of bonds

Note: “Real estate” and “Other” were combined as “Alternatives” to depict a trend in non-traditional asset classes. Asset allocations are plotted as the median allocations for each group.
Source: Office of the Superintendent of Financial Institutions pension fund dataLast observation: 2018

Reaching for resilience

At first sight, the fact that many plans switched from public equities to private markets suggests that they increased risk exposures to reach for higher yields. Moreover, anecdotal evidence suggests that many plans also increased their bond allocation to corporate bonds and made greater use of financial leverage.

However, the plans in our sample also shifted to a majority allocation in bonds, which have lower average returns than stocks. As a result, the long-term expected return of their overall portfolio decreased to 5 percent in 2018 (median across large shifters) from 5.25 percent for plans with a 60/40 portfolio (median across non-shifters). These patterns are inconsistent with a standard reach for yield.

Instead, we argue that the portfolio shift is better explained as a strategy to mitigate solvency risk. Because low interest rates have eroded solvency surpluses, plans have only a small buffer remaining to avoid a solvency deficit caused by further declines in the interest rates or the stock market. Without such a buffer, these market events may trigger a special cash payment from plan sponsors. The decline in interest rates has therefore increased solvency risk.

The portfolio shift toward bonds and alternative assets improves a pension plan portfolio’s resilience to solvency risk in three ways. First, a majority allocation to bonds allows the plan to hedge against fluctuations in solvency liabilities that come from fluctuations in market interest rates (Fisher and Weil 1971). Intuitively, both bond prices and solvency liabilities increase as market interest rates fall, so the bond allocation offsets the increase in liabilities. The use of financial leverage can also help the plan to further protect against interest rate risk by increasing duration.

Second, for plans that shifted their portfolio, investments in alternative assets and corporate bonds partly offset the lower returns from government bonds. The evidence in our sample suggests that the shift to alternatives relates directly to the shift to bonds: plans that do not increase their government bond allocation also do not increase their holdings of alternative assets.

Third, alternative assets that are not mark to market, such as private equity and real estate, also help reduce solvency risk. That is because investing in these assets reduces the volatility of asset values on the plan’s balance sheet (compared with public equities), which mitigates fluctuations in the solvency ratio. Additionally, investments in real assets, such as private infrastructure, tend to produce bond-like cash flows, which further protects the plan against fluctuations in the interest rate (Beath et al. 2021).

To assess whether the portfolio shift improves resilience to solvency risk, we plot the evolution of the median solvency ratio for each of the three groups along with the median solvency discount rate (Chart 4). We focus on the last five years of the sample (2013 to 2018) to ensure that asset allocations are distinct across groups.

Chart 4 shows that all plans started with a similar solvency ratio in 2013. However, a significant wedge between the solvency rate of shifters and non-shifters opened, which persisted until the end of our sample. Overall, plans that shifted were 30 percent less volatile than those that did not (as seen in the percent change in their solvency ratio).

Chart 4: Large shifter plans are more resilient to solvency risk

Source: Office of the Superintendent of Financial Institutions pension fund dataLast observation: 2018

However, greater resilience to solvency risk comes at the cost of a lower rate of return on the portfolio. This in turn leads to a greater valuation of going-concern liabilities and therefore a lower going-concern ratio. In a way, plans forgo some of the going-concern surplus to achieve greater resilience to solvency risk.

Chart 5 confirms that large shifter plans have lower going-concern ratios than moderate and non-shifter plans during this period. However, this trade-off is not binding because all groups maintain a going-concern surplus in 2018.

Chart 5: Large shifter plans forgo expected return but remain in a going-concern surplus

Source: Office of the Superintendent of Financial Institutions pension fund dataLast observation: 2018

In the Appendix, we present the results of a series of robustness tests and rule out alternative explanations of these patterns. We confirm that large shifter plans do not:

• return to solvency because they made special cash payments
• have low sensitivity to fluctuations in the interest rate because of a lower liability duration
• return to solvency because of the recent change in the solvency valuation method allowed under the OSFI regulations.

Together with the solvency requirement, recent accounting reforms may have contributed to some plans shifting their portfolios toward alternative assets and bonds. These reforms may explain why moderate shifters started later.

Passed in 2011, International Financial Reporting Standard (IFRS) 13 stipulates that employers report on their balance sheets the value of pension liabilities based on a long-term bond yield (similar to OSFI guidance for computing the solvency liabilities). This accounting reform magnifies the sponsors’ exposure to lower interest rates and therefore may have contributed to the portfolio shift.

Plans that shift are larger, more mature and manage assets in-house

If portfolio shifting increases resilience to solvency risk, why have some plans not shifted? We find that plan size is a key driver of shifting portfolios. In our sample, 85 percent of plans whose AUM exceeded $0.5 billion shifted their portfolio, whereas only 52 percent of plans with AUM below $0.5 billion did. Large plans have greater ability to invest in alternative asset classes than small plans do.

Not every large plan shifted, however. Among the 33 plans with AUM above $1 billion, 6 kept the standard 60/40 portfolio allocation. To identify additional drivers of the shift, we differentiate large shifter from non-shifter plans at the end of our sample in 2018 based on the proportion of:

• retired members in the plan
• stock and bond investments managed in-house

Chart 6 shows that the level of portfolio shift increases significantly for plans with more retired members and more in-house management.

Chart 6: Plans that shift are more mature and manage more of their assets in-house

Note: “Retired members” is the median proportion of retired plan members between 2016 and 2018. “In-house management” is the median proportion of assets managed internally, proxied as the proportion of stock and bond investments held directly.
Source: Office of the Superintendent of Financial Institutions pension fund dataLast observation: 2018

Chart 6 shows that large shifter plans are more mature: approximately 60 percent of their total members are retired as opposed to 35 percent for non-shifter plans. It is likely that large shifter plans have accrued greater pension liabilities. Therefore, potential deficits and special cash payments to restore solvency would have a larger impact on the sponsors’ business operations. So even though these more mature plans have shorter liability duration than younger plans, they have a greater need to manage solvency risk.

Large shifter plans also manage approximately 90 percent of their stock and bond investments directly, whereas non-shifter plans manage virtually none of these investments. Recent work by Beath et al. (2021) shows that Canadian pension plans that manage more of their assets in-house can save on costs and move resources toward managing risk. We reason that a robust risk management unit in large shifter plans is better at mitigating solvency risk.

Gang’s (2020) case study of Air Canada supports our findings. In 2009, Air Canada’s eight defined benefit pension plans faced a $2.6 billion solvency deficit while holding the traditional 60/40 portfolio, all managed externally. Air Canada’s eight pension plans were maturing quickly, with 60 percent of its liabilities tied to retired members, indicating a greater amount of retirement benefits coming due. Through the enhanced capabilities of in-house management, the plans gradually shifted to bonds and alternatives and used financial leverage. The plans achieved a solvency surplus by 2019, despite the decline in interest rates during that period.

Non-shifter plans face greater future solvency risk

What are the solvency risk implications for the 36 percent of pension plans that maintain a traditional 60/40 portfolio?

These plans may have good reasons for not shifting their portfolio. For example, they may not have the same ability to access private markets as large plans do. They may also be inclined to not shift if they have relatively small pension liabilities, as Chart 6 illustrates. Another reason is that many of these plans are currently running a solvency deficit. In this context, they may view the portfolio shift as expensive. Intuitively, when a plan falls below its solvency requirement, a shift to bonds that matches its liabilities will crystalize the deficit, requiring special cash payments to attain solvency.

However, the risk of keeping the 60/40 portfolio is that plans remain exposed to mark-to-market fluctuations in asset values and interest rates. Consider, for example, March 2020, when stock markets dropped by 37 percent and short-term interest rates fell by 1.5 percent. As a result of this exposure to solvency risk, plan sponsors should be ready to make special payments to the pension plans if the solvency deficits persist.

Conclusion

The secular decline of interest rates has resulted in:

• binding solvency regulations
• many Canadian, private, defined benefit pension plans running persistent solvency deficits

Among the 128 plans in our sample, 64 percent of plans responded by shifting their portfolio significantly from a traditional mix of 60 percent stocks and 40 percent bonds to a greater allocation in:

• alternative assets, such as private equity, real estate and infrastructure
• bonds

We argue that this portfolio strategy aims to reduce solvency risk while continuing to generate returns.

Our analysis provides a novel perspective into the reach-for-yield phenomenon. It supports the case that the risks associated with investments in alternative assets, such as corporate bonds, and other strategies, such as use of financial leverage, should not be studied in isolation. Instead, they should be considered as elements of the total portfolio strategy for managing solvency risk.

Moving forward, it will be interesting to investigate how reforms in solvency regulation—including the recent reforms adopted by provincially regulated plans in Canada—affect the overall portfolio strategy of pension plans.

References

1. Beath, A., S. Betermier, C. Flynn and Q. Spehner. 2021. “The Canadian Pension Fund Model: A Quantitative Portrait.” Journal of Portfolio Management Investment Models.
2. Bédard-Pagé, G., A. Demers, E. Tuer and M. Tremblay. 2016. “Large Canadian Public Pension Funds: A Financial System Perspective.” Bank of Canada Financial System Review (June): 33–38.
3. Cremers, M. and A. Petajisto. 2009. “How Active Is Your Fund Manager? A New Measure That Predicts Performance.” Review of Financial Studies 22 (9): 3329–3365.
4. Feunou, B. and J.-S. Fontaine. 2019. “The Secular Decline of Forecasted Interest Rates.” Bank of Canada Staff Analytical Note No. 2019-1.
5. Fisher, L. and R. L. Weil. 1971. “Coping with the Risk of Interest-Rate Fluctuations: Returns to Bondholders from Naive and Optimal Strategies.” Journal of Business 44 (4): 408–431.
6. Gang, Y. 2020. “Charting a New Course at the Air Canada Pension Plans.” Benefits Canada. May 15, 2020.
7. Greenwood, R. and A. Vissing-Jorgensen. “The Impact of Pensions and Insurance on Global Yield Curves.” 2019 Swiss Finance Institute Research Paper No 19-59.
8. “Taking Back Control.” 2020. The Economist. November 12.

Acknowledgments

We thank Benoit Brière, Annick Demers, Tamara DeMos, Jean-Philippe Dion, Toni Gravelle, Grahame Johnson, Anne-Marie Lainesse, Stephen Murchison, Jean-Claude Primeau, Martin Robichaud, Robert Scott, Virginie Traclet and Adrian Walton for insightful comments and suggestions.

Appendix

The Office of the Superintendent of Financial Institutions (OSFI) provided anonymized data. Federally regulated Canadian private pension plans are required to submit three forms to OSFI each year:

• OSFI 49 (Summary)
• OSFI 60 (Statement of Net Assets)
• T1200 (Actuarial Information Summary)

Our sample contains high-quality information on the asset and liability portfolios of all federally regulated private plans over the past 20 years. We analyze the information provided in these forms and focus on the 128 defined benefit and hybrid plans that:

• were operating in 2018
• manage more than $50 million in assets

Table A-1 summarizes the size of the plans by members and total assets.

Table A-1: The OSFI data provide information about a diverse group of small and large plans

Portfolio shift metric

To determine how pension plans have adapted their portfolio, we define a portfolio shift metric that measures a plan’s change in asset allocation between times t₁ and t₂:

$$Portfolio\,Shift = \frac{1}{2} \sum_{i=1}^4 \bigl| w_{t_{1},i} - w_{t_{2},i} \bigl|, $$

where w_t2,i and w_t1,i are the portfolio weights for asset class i at times t₁ and t₂. The sum is indexed over four asset classes:

• stocks (public equities, stock mutual funds)
• bonds (cash, short-term notes, long-term bonds)
• real estate (public and private)
• others (private equity, infrastructure)

The portfolio shift metric is bounded by 0 and 1. A value of 0 indicates that asset allocation has not changed between t₁ and t₂. By contrast, a value of 1 indicates a complete overhaul of the portfolio. The shift metric is inspired by the active share metric used in the mutual fund literature to quantify the deviation of a mutual fund portfolio from a benchmark portfolio (Cremers and Petajisto 2009).

To remove the effects of transitory portfolio fluctuations arising from market movements, we measure the 2004 portfolio allocation as the median allocation between 2004, 2005 and 2006. Likewise, we measure the 2018 allocation as the median allocation between 2016, 2017 and 2018.

Lower volatility in percent change of solvency ratio

To assess whether the portfolio shift has improved resilience to solvency risk, we calculate for each plan the yearly percent change in the solvency ratio from 2014 to 2018. We then calculate the standard deviation of these five observations. For non-shifters, the median standard deviation is 5.51 percent. For large-shifters, the median standard deviation is 3.82 percent, which is approximately 30 percent lower than that of non-shifters.

Alternative explanations for the improved resilience of large shifter plans to solvency risk

We run a series of robustness tests to ensure that the improved resilience of large shifter plans to solvency risk is not driven by alternative explanations.

The first alternative explanation could be that large shifters received significant special cash payments between 2013 and 2018. To verify this explanation, we calculate the cumulative special payments during this period for the three groups as a proportion of their total solvency liabilities in 2018.

Chart A-1: Large shifter plans pay lower special cash contributions but restore solvency

Source: Office of the Superintendent of Financial Institutions pension fund dataLast observation: 2018

Chart A-1 shows that non-shifter plans made special payments amounting to almost 8 percent of their solvency liabilities over five years, whereas large shifter plans paid cash payments of 2 percent of their solvency liabilities. Therefore, special payments do not restore solvency for large shifter plans.

The second alternative explanation is that large shifters have lower liability duration and therefore less sensitivity to interest rate changes, irrespective of their asset mix. To test this explanation, we separate young and mature plans depending on whether more than 50 percent of their members are retired in 2018. We then compare the median standard deviation in the yearly percentage change of the solvency ratio between 2014 and 2018 across groups.

Chart A-2: Mature pension plans that shifted portfolios do not have more volatile solvency surplus

Source: Office of the Superintendent of Financial Institutions pension fund dataLast observation: 2018

Chart A-2 shows that large shifter plans have a lower standard deviation of their solvency surplus than non-shifter plans irrespective of their maturity profile. The results are most pronounced for younger plans. The large portfolio shift reduces the volatility of solvency surpluses for young plans, but not for mature ones.

The third alternative explanation could be that large shifter plans recently switched to a different method for estimating the solvency discount rate (OSFI’s replication portfolio method). Chart A-3 plots the median solvency discount rate for non-, moderate and large shifter plans throughout the sample period. We find no differences across groups. Therefore, changes in solvency liability valuation do not restore solvency for large shifter plans.

Chart A-3: The median solvency rate is similar for non-, moderate and large shifter plans

Source: Office of the Superintendent of Financial Institutions pension fund dataLast observation: 2018

Additional robustness tests

Certain pension plans invested a significant proportion of their assets in balanced and segregated mutual funds, without further details on fund holdings. For our analysis, we assume these mutual funds are invested in a traditional mix of 60 percent stocks and 40 percent bonds and add the respective proportions into direct stock and bond asset allocations. To verify this assumption, we identify the plans that invest a significant proportion in balanced and segregated mutual funds. We omit these plans and find that it does not alter our results and the remaining sample largely follows a traditional 60/40 allocation.

A few plans have identical asset allocations and shifts over time and therefore appear to be operated by the same sponsor fund. As a robustness exercise, we merge these sibling plans into unified plans and find that it does not alter our results.

Finally, we want to ensure that our results are not driven by the shift to bonds as a result of termination of plans. We compare the portfolio shift observed in our data to the shift undertaken by a pension plan in the years following its announcement of termination. Plans that are terminating typically unwind using a rapid (less than two years) shift to allocate 100 percent of their portfolio to fixed-income securities. This shift is inconsistent with the gradual shifts observed in our data.