Facing the Future: What Works for Retirement Saving and What Doesn’t

In this, our final article looking at how different plan designs/investment strategies worked over the last five years, we want to synthesize what we have learned in a way that will help sponsor decision-makers. We’re not going to predict the future – our assumption is that the market is better at that than we are. Rather, we are going to lay out our view of what things (generally) work in retirement finance, in both DB and DC plans. And when they don’t work.

In the main body of the article, we detail our theses about retirement finance generally (the things that work) and then discuss counterexamples (when those things don’t work).

We begin, however, with three charts.

A Brief Summary of the Analysis

For DC Participants

(We discuss this chart in detail in our article How DC Participants Are Doing, in Four Charts)

These are two age-55 participants who start out (at the end of 2020) exactly the same. Each has an account balance that buys them a $1,000 per month annuity. One buys that annuity, the other invests in the plan’s target date fund. The green line shows how much real retirement income (in 2020 dollars) the non-annuitizing participant can buy at the end of our period (end of 2025). The orange line shows how much the annuitizing participant’s 2020 annuity is, after five years of inflation, now worth.

The obvious point – participants who stayed in the market rather than annuitizing did twice as well, in “retirement income buying power,” than participants who annuitized. Why is that?

For DB Participants

This chart shows similar data for three different types of DB plans – traditional DB, “traditional” cash balance (30-year Treasury crediting rate), and market-return cash balance (crediting rate based on target date fund returns).

(We discuss this chart in more detail in our second article in this series – How are DB participants doing?)

Again, all participants start out with a benefit worth the same amount at the end of 2020. Participants in the market-based cash balance plan, like the DC participant who “stayed in the market,” did significantly better, in terms of how much real retirement income their account can produce. Interestingly, the target date fund in the market-based cash balance plan grew (in terms of retirement income producible) at a steadier (less volatile) rate than the “traditional” cash balance plan with the “fixed” income crediting rate. And the MBCB ultimately produced more retirement income. Why is that?

For DB Sponsors

Sponsors of DC and market-based cash balance plans, of course, carry no risk. But sponsors of traditional DB plans and “traditional” cash balance plans do carry risk. As we discuss further below, traditional DB plan sponsor risk can be hedged (via liability-driven investments). The traditional cash balance plan sponsor risk cannot. This chart shows the sponsor “performance” (in terms of funding surplus or shortfall) of the “risk-bearing” DB plans: traditional DB plans where the sponsor does not hedge but rather seeks returns; traditional cash balance plan sponsors that explicitly seek returns (called “arbitrage”); and sponsors of TCBs that try to “hedge” (by buying fixed income assets).

(We discuss this chart in detail in our article How are DB sponsors doing?)

The reader will notice that the gains made by return-seeking sponsors of traditional DB plans come at the expense of participant losses (shown in the second chart above). And the traditional cash balance plans, which were supposed to tame the ongoing risk/volatility that drove most sponsors out of the traditional DB plan business, produced more volatility for both participants and sponsors. Why (again) is that?

Our Answers to Those Three Questions, in Brief

Retirement Finance in Five Bullets

Asset values and interest rates (standing in for the cost of retirement income) are inversely correlated. When interest rates go up, asset values (generally) go down, but the amount of retirement income those assets can buy goes up. DB managers are familiar with this tradeoff, and the implementation of liability-driven investment strategies (LDI) exploits it to hedge liabilities. This relationship between asset values and interest rates is more obscure to DC participants, who are generally more focused on account balance than on retirement income. Expected inflation is built into market interest rates. Unexpected inflation is, however, an uncompensated reduction in the buying power of nominal income. These variables/correlations produce the following results.

• In the “normal situation” (a normal year), when interest rates go down, asset values go up and vice versa (e.g., 2014, 2022).

• In a “good year,” any decline in asset values resulting from an increase in interest rates is less than the gain in retirement income from that increase. (Or vice versa.) This would be a “real” bull market (e.g., 1999, 2009).

• In a “bad year,” the decline in asset values resulting from an increase in interest rates is greater than the gain in retirement income from that increase. (Or vice versa.) This would look like a recession (e.g., 2000, 2008).

• Expected inflation should wash through all this, as already reflected in interest rates.

• Unexpected inflation will generally punish current holders of fixed-income assets (most of all, annuitants). Its effect on holders of equities is more ambiguous and depends in part on portfolio companies’ ability to adapt to/compensate for inflation.

What Works

Given the correlations between these risks, we believe that the most robust strategy is to put them all in one place – with the participant. This is what happens in a 401(k)/DC plan and in a market-based cash balance plan.

In a (utopian) environment where unexpected inflation is not a risk, the simplest retirement investment strategy would be to invest in high-quality fixed-income assets that would produce the income a participant will need in retirement. Something like interest rate “strips” producing income over the participant’s years, age 65-85.

But, in the context of unexpected inflation (very much present over the 2021-2025 period we have looked at), a target date fund strategy produces much better results. Fixed income assets generally (and annuities in particular) are uniquely vulnerable to unexpected inflation shocks.

And because, practically, unexpected inflation risk is (in all current plan design options) always on the participant, the most robust solution to these challenges is (again) to leave this bundle of risks with the participant. Doing so allows the participant to exploit key correlations to actually reduce volatility in his long-term project of producing adequate income in retirement. That’s it in a nutshell. Leave all financial risks with the participant (401(k)/DC or market-based cash balance) and put the participant in a target date fund. And yes, that sounds counterintuitive, especially for sponsors that live in an all account-based (401(k)/DC or market cash balance) world. Why that is the case is what the rest of this article is about.

That’s it in a nutshell. Leave all financial risks with the participant (401(k)/DC or market-based cash balance) and put the participant in a target date fund. 

And yes, that sounds counterintuitive, especially for sponsors that live in an all account-based (401(k)/DC or market cash balance) world.  

Why that is the case is what the rest of this article is about. In what follows, we’ll unpack these conclusions – why we think they “work” – and consider situations in which they don’t. 

*     *     * 

Theses

1. Retirement savings (always) present four risks: asset performance; interest rates; inflation; and mortality/longevity.

These risks flow from the fundamental task of “saving for retirement” – the saver must deploy income today (generally, savings from wages) to secure a stream of goods and services during retirement (for younger individuals, a quite distant future). 

• Asset performance risk is, under modern conditions (efficient capital markets, efficient asset managers, e.g., the mutual fund industry), generally regarded as an opportunity. 

• Interest rates are generally understood as representing something like the time value of money. But they are (literally and in fact) the market “price” for turning assets into income – that is, for turning a pile of money in the saver’s account into a stream of income over time. There are (many) markets for this risk, and it is relatively easy to hedge it (generally at the cost of giving up the opportunity provided by asset performance risk). 

• Inflation is something of a wild card. Expected inflation is generally reflected in interest rates. The problem is unexpected inflation, the incidence of which is (by definition) unpredictable. In many contexts (e.g., low and predictable inflation from 2009-2021), inflation is nearly transparent. Then, e.g., after some policy or fiscal shock, the saver/investor sees inflation take away, say, 10% of the buying power of her savings. Pre-inflation, she thought she’d need $3,000 a month to retire on; she now needs $3,300 a month. The biggest problem inflation risk presents for retirement savings is that there is (effectively) no way to hedge it, unless you count Treasury Inflation-Protected Securities (TIPS), which are generally regarded as prohibitively expensive. 

• Mortality/longevity – how long the individual’s retirement income will have to last – is for the typical individual a roll of the dice. But this risk is – with a big enough population (say, 500 lives) – totally hedgeable through risk pooling. The only tradeoff is that the participant gives up the asset performance opportunity. Among the risks discussed, mortality risk can be uniquely ameliorated by pooling. 

Sidebar – Why interest rates and inflation matter.
Especially for sponsors that operate in an account-based plan (401(k) and cash balance plans) world, it may not be intuitive why anything other than account balance and mortality (longevity risk) matters. There are several reasons why these two risks (interest rates and inflation) matter, and why they present special challenges for the retirement saver. 

1. The price of financial assets themselves is based on the income an investment will produce over time, determined by a discount rate that is derived from interest rates and that changes when interest rates change. This is explicitly the case for bonds but is also implicitly true for equities (via assumptions about a stock’s price-earnings ratio). Obviously, some market participants (e.g., traders) are only concerned about (current) asset values. But they still focus very much on interest rates/interest rate policy. Because when interest rates go down, asset values generally go up, and vice versa. 

2. Unlike traders, retirement savers require income in retirement. And, as we say in the main body of the article, interest rates define how much income you can get out of a given asset pile. So, if (in a non-typical year) interest rates go up, but the value of your asset pile stays the same, then you can produce more retirement income, even though your account balance hasn’t changed. 

3. Unexpected inflation drives interest rates (in a somewhat “noisy” way). Expected inflation is generally baked into current interest rates. 

4. Inflation reduces buying power. While expected inflation can be accounted for in the participant’s retirement planning, it nevertheless reduces the buying power of any fixed-income asset. For publicly traded assets (stocks and bonds), the effect will generally show up in asset values – but (and this is critical for, e.g., traditional DB benefits) it does not show up in annuity values. Rather, it is simply the case that any inflation reduces the value of an annuity (in terms of real goods and services). 

5. Unexpected inflation – like we saw in 2022 – is a loss to all holders of fixed income assets, most of all annuitants. 

6. Interest rates and inflation are correlated in very significant ways, with each other and with asset performance, and any analysis of retirement savings strategies/performance must account for those correlations. 

2. During the accumulation phase (that is, pre-retirement), taking these risks as a bundle, equities (and other equity-like investments) provide the best “tool” for turning present wages/assets into a stream of future goods and services.

As our previous articles have shown, the first three of these risks – the “financial” risks of asset performance, interest rates, and inflation – are (typically) highly correlated. (See the Sidebar on “Why interest rates and inflation matter.”) Specifically, when interest rates go down (increasing the cost of future income), asset values typically go up. Which is why the stock market loves low interest rates. 

• Generally, fixed-income investments provide a risk-free solution in an inflation-free world. And if there were no inflation, hedging would be as simple as the liability-driven investment advocates tell you: all you have to do is buy bonds that provide the future income stream you need (or buy an annuity, and then the annuity carrier will buy those bonds). 

• But fixed-income investments are uniquely vulnerable to inflation risk. While a portfolio of bonds can secure a stream of future income, inflation can derail the ultimate objective, i.e., the amount of goods and services that can be bought with that income. 

Sidebar – Why retirement income (not account balance) matters.
It’s not primarily due to longevity risk. It’s very simple. Have a conversation with an individual approaching retirement, and they’ll tell you: once you stop working and there’s no longer a weekly paycheck, your primary concern is whether your income from savings will meet your ongoing expenses. 

That is why in this series, our endpoint – our metric for retirement success – is income, not account balance. Account balance is still a key variable, but the ultimate question is: how much income will that account balance produce? And the answer to that question depends on interest rates. 

3. Thus, plan designs that put the three financial risks in one place provide the best result.

There are exactly two designs that do this: 401(k)/DC plans and market-based cash balance plans leave the three financial risks with the participant. These “true” account-based plans generally default to something like the target date fund investment strategy we have discussed in our prior articles. And they did best for the participant, while zeroing out sponsor risk. 

4. The other plan designs – traditional defined benefit (TDB) and traditional cash balance (TCB) plans – create unmanageable risks for participants and create perverse incentives for sponsors. 

• Traditional defined benefit plans allocate asset performance and interest rate risk to the sponsor but leave the participant with inflation risk (post-retirement at least, but pre-retirement too for frozen or non-indexed benefits). In these plans, a great year for the sponsor – say, one in which there is a significant, inflation-driven increase in interest rates, dramatically reducing liabilities – may correspond to a bad year for a participant, as she loses 10% of the value of her pension to inflation. 

• Traditional cash balance plans may mitigate this effect somewhat, as under these plans, while the employer carries asset performance risk, the participant holds both interest rate risk (e.g., via the interest rate-driven cost of buying an annuity) and inflation risk. When interest rates decline (as they mostly did from 1982 to 2020), TCB participants suffer from the higher cost of retirement income but don’t benefit from higher asset returns for stocks and bonds. Conversely, as the charts above show, when interest rates rise (as they did sharply in 2022), participants enjoy cheaper retirement income without suffering from lower asset prices. In the earlier period, sponsors enjoyed lower costs, but during 2022, TCB sponsors uniquely suffered. (We discuss these issues in detail in our article How are DB Sponsors Doing?) 

• In defined benefit plans, sponsor outcomes depend largely on investment strategy (return seeking vs. hedging), with sponsor gains correlated with participant losses. In effect, equity gains/returns associated with inflation go to the sponsor, while inflation losses go to the participant. In this context, hedging (e.g., LDI in a traditional DB plan) looks like the worst strategy, as the participant still sustains the loss-to-inflation, but the sponsor foregoes the returns available from stocks that would better hedge against inflation. 
  

5. During the distribution phase, annuities provided through DB plans enjoy much better pricing than commercial annuities.

Unlike financial risks, mortality risks can be ameliorated by pooling the risk over hundreds of lives, and employers are in a unique legal position to provide such pooling on behalf of their participants. This is the basis for the 15-30% “DB annuity advantage” we discuss in our article The DB retirement income advantage. 

Anti-theses 

We believe the above theses generally hold in normal conditions and produce exceptional results in good conditions. When do they do poorly? 

• When stocks and interest rates fall simultaneously, typically in a recession. The bursting of the dot-com bubble during 2000-2002 is a good example. In these conditions, fixed-income assets outperform return-seeking assets like stocks. 

• Under conditions of very low inflation, bonds/annuities represent a legitimately safe retirement savings investment – that is, the higher premium for equity investments is simply compensation for a higher beta. As noted, “expected inflation” is generally already accounted for in current interest rates. The critical problem is unexpected inflation, and that always remains a risk. 

• Immediately after a “once in a generation” market shock. In other words, in a crash, (high-quality) bonds are safer than stocks. 2008 is a good example of this. 

These are all reasons why the entity/individual holding asset performance and (with respect to traditional DB plans) interest rate risk might want to hold bonds (and even, conceivably, hedge those risks out). They are not, as in DB designs other than market-based cash balance plans, a good reason to separate inflation risk from those other risks. 

Implications of These Conclusions 

In addition to the implications of these conclusions for plan design, they are particularly pertinent to the current debate over “retirement income” and annuities in DC plans. Take another look at the first chart above, comparing the retirement income of a participant who didn’t annuitize at the end of 2020 to one who did. The result for early annuitization is not pretty. 

We will continue to follow this issue. 

*     *     * 

This is a publication of O3 Plan Advisory Services. If you have any comments or have questions about regulatory developments, please contact your relationship manager or Mike Barry at mbarry@octoberthree.com.  

The information, analyses, and opinions set out herein are for general information only and are not intended to provide specific advice or recommendations for any individual or entity. Nothing herein constitutes or should be construed as a legal opinion or advice. You should consult your own attorney, accountant, financial or tax advisor, or other planner or consultant with regard to your own situation or that of any entity which you represent or advise. 

Information set out or referred to above has been obtained from sources believed to be reliable. However, neither O3 Plan Advisory Services nor any of its affiliates has verified the accuracy or completeness of any such information. All information is provided “as is,” and O3 Plan Advisory Services and its affiliates expressly disclaim all express and implied warranties regarding the information. Neither O3 Plan Advisory Services nor any of its affiliates shall have any liability for any use of the information set out or referred to herein.