Featured Savings Rates
Featured Accounts

Are CD Investors More Likely To Have Sustainable Withdrawals?

Bozo   |     |   762 posts since 2011

The popular financial press oft notes the 4% "SWR" in retirement. Economists study it. PhD candidates write volumes deriding it. Yet one thing strikes me as odd. To my knowledge, nobody has ever examined the 4% sustainable withdrawal rate (SWR) in the context of a retiree with 50% equities and 50% in a CD ladder. Herewith a challenge to any PhD candidate: crunch the numbers.

Background: Going all the way back to the granddaddy of SWR analyses (the Trinity Study), the underlying assumption was that the fixed-income component of any portfolio would be in bond funds. In a falling-rate environment, which existed from roughly 1980 to just recently, it could reasonably be assumed bond funds would do "OK". In a rising-rate environment, bond funds are problematic, which has led some wags to opine that the 4% SWR is no longer viable. They opine, in effect, that bond funds may well be dead money.

Issue: Is a CD ladder of 5-yr CDs better than a bond fund for calculating SWR, and should academics focus on such a ladder for their analyses?

Analysis: Most economists (aside from those in the White House) forecast roughly 2% growth as far as the eye can see, Trumponomics or not. Dividends are showing a steady and healthy 2% (on average). DA readers are well aware that obtaining 2%+ on a 5-yr CD is not terribly difficult (indeed, it seems to be the "going rate" for some 2-yr CDs). CDs do not have the potential "drag" on yield created by a rising-rate environment. The NAV of a retail CD does not decrease when yields increase, unlike bond funds.

Challenge: OK, all you PhD candidates, go back to your computers and algorithms, and plug in all your assumptions, but make one wildly different assumption. Assume the fixed-income portion of a retiree's portfolio is in a ladder of 5-yr CDs which will never be impacted by an interest-rate rise (other than being beneficial), and that the effective rate is 2%. Throw out all those old, tired, assumptions about bond funds as the only vehicle for fixed-income in a balanced portfolio. Is a person more likely to sustain a 4% withdrawal rate with the afore-mentioned CD ladder, or a bond fund?

anitje   |     |   1 posts since 2017
I re-posted this on bogleheads.org and received a series of responses that might be of interest to you.
Check this link: https://www.bogleheads.org/forum/viewtopic.php?f=10&t=220523
Bozo   |     |   762 posts since 2011
anitje, I could not open.
Ally6770   |     |   2,317 posts since 2010
Did you copy and paste? I had no problem opening the link. We had never purchased a bond. Actually have never cashed in any CD's except when we purchased our last home but they had matured and prices on new homes were low after the crash and were still low in 2010 when we finally found a house that we wanted.
Bozo   |     |   762 posts since 2011
Was finally able to open the link. The responses were fairly predictable for Bogleheads. Bond fund advocates thought I was nuts, CD advocates thought not.

The one thing I missed was anyone responding to my challenge. Plugging all those assumptions into your computer, with one modest change (using a CD ladder rather than bond funds), which scenario fares better for purposes of a sustained withdrawal rate of 4%? Maybe I missed it.

Let's face reality. CD ladders aren't sexy. They don't exactly get PhD candidates' blood boiling, or examiners at PhD orals overly enthused. As a result, a person who focuses on CD ladders in a PhD thesis is likely to be rewarded with a resounding "thud". Not exactly an employment-enhancing result towards your PhD. 
KevinM   |     |   26 posts since 2012

Your characterization of the Bogleheads discussion does not do it justice.

First, with about 75% of my fixed income in direct CDs (and 70% of portfolio in fixed income), I fall into the category of "CD advocate"--but I'd clarify to call it "direct CD advocate". You'd also get that if you scanned the close to 2,000 CD posts of mine on the BH forum. While I don't think you're nuts, I do think you may not fully understand how similar CD ladders, bond ladders, and bond funds are, which I addressed briefly in the BH thread. But before getting into that, here are a few key points from the BH thread that your post here stimulated.

A major point is that in a 50/50 stock/fixed-income portfolio, the fixed-income portion has a relatively minor effect on the SWR, since it's the stock returns that dominate the portfolio returns. This was explored a bit by one poster, AlohaJoe, by comparing the effect of short-term vs. intermediate-term bonds on the SWR, noting that short-term bonds would be a better proxy for a 5-year CD ladder (with an average maturity of 2.5 years). AlohaJoe's analysis of this was in response to some pushback by me to several of the points that he made earlier in the thread, in which I basically showed that short-term fixed income fared much better in real terms during the 1970s when nominal rates were increasing and inflation was high.

I subsequently showed that the shortest term fixed income (cash--no term risk) resulted in significantly higher terminal value than 10-year Treasuries when used in a 50/50 portfolio for 1972-1981, even better in a 100% fixed-income portfolio, and best of all in a 30/70 stock/fixed portfolio. But SWR studies typically look at 30-year periods, not 10-year periods--I was mostly demonstrating that keeping term risk low during periods of rising rates and high inflation can be beneficial to a portfolio.

The next important point is that your understanding that the value of a 5-year CD ladder is unaffected by rising interest rates is incorrect, especially if you don't specify whether you're talking about brokered CDs or direct CDs. You did mention "retail CDs" in the previous paragraph, so perhaps you were thinking about direct CDs in your ladder comment.

Changing interest rates will have a similar impact on the market value of a 5-year ladder of brokered CDs as on a 5-year ladder of Treasuries, and this will be similar to the impact on a bond fund with an average maturity of about 2.5 years.

Two things are likely to be different about the CD ladder: 1) Assuming an environment like we've had over the last 6-7 years (at least), the yield of the CD ladder may be 50-100 basis points higher than the Treasury ladder; 2) It will cost you more to sell a brokered CD than to sell a Treasury, as you would likely have to do to rebalance when stocks decline significantly; the bid/ask spread on brokered CDs is much higher than that for Treasuries.

I think it's likely that the higher yield of the CDs will more than compensate for the higher cost of selling CDs to rebalance, but there's not a lot of easy visibility into bid/ask spreads for brokered CDs, and it will depend on your rebalancing policy.

Of course direct CDs with low early withdrawal penalties (EWPs) have the significant benefit of paying a much smaller price to "sell" the CD after rates have risen (of course for this benefit, you give up the potential of being able to sell your fixed income at a higher price to rebalance into stocks in a scenario like late 2008, when stocks tanked and Treasuries increased in value). At any rate, because of the EWP it's not strictly true that there's no impact on value when rates increase, since to take advantage of higher rates you must do an early withdrawal and pay the EWP.

So, although I'm not running an actual model, we can think through how a 5-year direct CD ladder would compare to a 5-year Treasury ladder, which again, would be similar to a short-term Treasury (bond) fund. As mentioned above, you'll have a higher yield from the CD ladder (my average yield premium over the last 6+ years is about 115 basis points), and you'll pay an EWP to rebalance into stocks if your investment policy and market movements require it. I think the benefit of the higher yield is likely to outweigh the downside of paying the EWP, and assuming banks and CUs don't clamp down further on early withdrawal terms in a sustained rising rate environment like the 1970s, the CD ladder return should be even higher than the original APYs.

AlohaJoe's analyis indicated that going with short-term fixed income in preference to intermediate-term fixed income increased SWRs by about 5% in the time period during which I showed that sticking with short-term increased cumulative returns by 3X. And if you expand the time period to the longest available in Portfolio Visualizer, Jan 1972-May 2017, the 50/50 portfolio with 10-year Treasuries did vastly better than sticking with cash for fixed income, and 100% stocks did by far the best of all.

With good direct 5-year CDs we essentially get intermediate-term yields with short-term risk (assuming low EWPs that are honored when we need them), so I'd estimate that using a good 5-year CD ladder for fixed income in a 50/50 portfolio might give us something a little better than using an intermediate-term bond fund, especially if we stick with no-credit-risk Treasuries for our bonds. But, over most 30-year historical periods for U.S. markets, stocks will dominate the returns of a 50/50 portfolio, and messing around at the margins with the fixed income is unlikley to have huge impact.

Good direct CDs will have a larger impact for those of us with lower stock allocations. Since my stock/fixed allocation is 30/70, direct CDs are likely to be more beneficial than they are likely to be for someone with a 50/50 AA, especially in a rising rate environment.

Bozo   |     |   762 posts since 2011
Kevin, I was, indeed, referring to "direct" retail CDs. What concerns me is that, in a rising-rate environment, having bond funds for one's FI allocation might entail more than "messing around at the margins".

For example, are you aware of any recent studies along the line I suggested, i.e., where the effective yield on an intermediate-term bond fund in a rising-rate environment is zero for as far as the eye can see? As the yield on the Ten was going ballistic (at least until mid-March), and the NAV on VBTLX was falling faster than a failed North Korean missile, well, you get my drift.

The point of my challenge was to suggest re-thinking the entire idea of intermediate-term bond funds (such as VBTLX) in analyses of sustainable withdrawal rates in a rising-rate environment.
Ally6770   |     |   2,317 posts since 2010
No matter which you choose you are not going to be 100% right 100% of the time. I prefer to have FDIC insurance. Bonds are never insured except for criminality. Whether it is a corporate bond fund or a muni you are at the mercy of the the bond holder (payer) and interest rates. I want to know what I have at all times, know that it is 100% insured and when and if I wanted or needed the money I know how much I would get.
KevinM   |     |   26 posts since 2012
Bozo said 'What concerns me is that, in a rising-rate environment, having bond funds for one's FI allocation might entail more than "messing around at the margins".'

For sure--direct CDs with low EWPs for the fixed-income component will provide a superior result in a rising-rate environment. Even cash will be better than intermediate-term bonds, as I pointed out for the 1972-1981 rising rate / high inflation environment.

My intuition was that over 30-year periods, stock returns dominate fixed-income returns to the extent that the type of fixed-income in a 50/50 portfolio wouldn't have a significant impact, but let's look at some data for a generally rising-rate environment. This won't consider portfolio withdrawals, but just portfolios rebalanced annually.

We'll look at a 30-year period of generally rising rates--1952-1981 (inclusive). Here are some real cumlative returns, with real annualized returns in parentheses:

-21% (-0.79%) - ITT (intermediate-term Treasuries)
+15% (+0.48%) - Tbill (3-month Treasury bill)
+121% (+2.67%) - 50/50 US stocks / ITT
+164% (+3.29%) - 50/50 US stocks / Tbill
+369% (+5.28%) - US stocks

So as is typically the case, over this 30-year period, U.S. stocks had much higher returns than intermediate-term U.S. bonds (which had a negative real return over this particular period). And we do see that stocks dominated the portfolio returns, but perhaps "messing around at the margins" understates the postive impact of keeping duration short during a 30-year period of generally rising rates.

And of course good direct CDs would do even better than Tbills if we were able to get the same yield premiums and low EWPs we've been getting over the last 6-7 years, and assuming that banks and CUs honored the early withdrawal requests (all of these assumptions are somewhat questionable).

Bozo said, "For example, are you aware of any recent studies along the line I suggested, i.e., where the effective yield on an intermediate-term bond fund in a rising-rate environment is zero for as far as the eye can see?"

I had already mentioned that I did my own "study" for the period 1972-1981, and yes, it showed that keeping term risk low in such an environment is beneficial for a 50/50 portfolio, and even more beneficial for a 0/100 portfolio. There's no argument here. The data shared above is more evidence that keeping term risk low in a longer-term rising-rate environment also is beneficial.

Bozo said, "As the yield on the Ten was going ballistic (at least until mid-March), and the NAV on VBTLX was falling faster than a failed North Korean missile, well, you get my drift."

Funny. But this was a very short time period, and it wasn't as bad as you indicate. It was mainly November 2016, after the election, when bonds took a big hit (VBTLX -2.64%), and VBTLX has been pretty much steadily recovering since then. And again, stocks did really well in November (VTSMX +4.44%), and over the last year have far outperformed bonds or CDs, so stocks have dominated the return of a 50/50 portfolio regardless of the fixed-income component.

For the one-year period ending May 31, here are the nominal returns:

17.53% - VTSMX (US stocks)
09.36% - 50/50 VTSMX/VBTLX
08.75% - 50/50 VTSMX/Cash
01.46% - VBTLX (intermedidate-term US bonds)

A 2% return for CDs would have added about 50 basis points for the fixed-income component relative to VBTLX, so about 25 basis points to a 50/50 portfolio. This is what I'd call messing around at the margins. I wouldn't use this short period to make my point.

Bozo said, "The point of my challenge was to suggest re-thinking the entire idea of intermediate-term bond funds (such as VBTLX) in analyses of sustainable withdrawal rates in a rising-rate environment."

Yes, as I said in my prior post, direct CDs with low EWPs definitely will help a 50/50 portfolio in a rising rate environment relative to an intermediate-term Treasury fund (adding credit risk to the bonds adds another dimension of risk which could help or hurt). Even cash does better than intermediate-term bonds in a rising rate scenario, and we've now seen more evidence to support this conclusion.

KevinM   |     |   26 posts since 2012
Since my previous reply, I've discovered that a "sustainable spending rate" metric has been added to the Bogleheads "Simba" backtest spreadsheet, which you can find with this Google search: simba site:bogleheads.org. It turns out that although keeping duration short had a significant positive impact on an annually rebalanced 50/50 portfolio during the 30-year period 1952-1981, the impact on SWR was minimal. Here are the SWRs for the portfolios of interest:

3.60% - ITT (intermediate-term Treasuries)
3.74% - Tbill (3-month Treasury bill)
6.65% - 50/50 US stocks / ITT
6.70% - 50/50 US stocks / Tbill
10.04% (+5.28%) - US stocks

So although there was a 69 point spread between the annualized real returns of ITT and Tbills (see my previous reply), there was only a 14 point spread between the SWRs for 100% ITT vs. 100% Tbills. And the 62 point spread between the annualized real returns of 50/50 stocks/bonds using ITT vs. Tbills translated to only a 5 point spread in the SWRs. And of course it was increasing stock allocation that had by far the largest impact on SWR.

I find it very interesting that the impact of keeping duration short has had much less benefit to SWR than to returns of a portfolio without withdrawals.

Again, direct CDs should do somewhat better than Tbills given the assumptions stated in my previous reply, but the results shown above indicate that the impact may not be very large on SWRs.

I added this personal note in the BH post I just submitted on this topic:

On a personal note, since I'm in the "safety first" camp as opposed to the "probability-based" camp, for now I'm sticking with my 30/70 stock/fixed portfolio, with 75% of the fixed income currently in direct CDs. The fixed-income portion of my portfolio is more than enough to constitute a liability-matching portfolio for me, so I simply don't need to take the risk of a higher equity allocation, and for me, low need to take risk trumps higher ability to take risk, especially factoring in my willingness to take risk (which I learned a lot about in 2008/2009, having retired in 2007).

Bozo   |     |   762 posts since 2011
Kevin, as always, thank you for the insights. I think your observations are quite lucid, and add much to the discussion here at Ken's blog.

I guess my takeaway is as follows: in a rising-rate environment, a lower-risk fixed-income investment option (such as a ladder of CDs) will improve sustainability of withdrawals, but not so much as one might suspect. I now feel much less guilt about plopping my maturing IRA CD into that 2% 2-year CD.

I did have to chuckle at the Ten. After bottoming-out at 2.14%, it has rebounded to 2.2%+.

Nice to chat with you.