CD 5-Year Report Card

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
User avatar
Kevin M
Posts: 10002
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

CD 5-Year Report Card

Post by Kevin M » Mon Nov 02, 2015 10:52 pm

I've just published part 3 of a 3-part blog-post series on the 5-year return of a 5-year direct CD I bought about five years ago compared to the 5-year return of a 5-year Treasury note I could have bought on the same date, and to the 5-year returns of various Vanguard bond funds.

CD 5-Year Report Card: Part 1 compares the returns of the CD and the Treasury note. This post is relatively short and simple.

CD 5-Year Report Card: Part 2 (Risk and Return) is pretty much just a discussion of the risk and return of fixed income investments. I felt I needed to cover this material as background to help with a rational comparison of the returns. This post is quite long and theoretical. It will be old hat to those who have a solid understanding of the risks and returns of fixed-income investments, but for others it may provide some useful background.

CD 5-Year Report Card: Part 3 is the main course, in which the returns are presented and discussed. It also is quite long, and is not light reading for those unfamiliar with analyzing the risks and returns of fixed-income investments. Hopefully the information in Parts 2 and 3 together will elucidate the points I'm trying to get across--at least to the interested and diligent reader.

Looking forward to comments, questions and discussion.

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

mur44
Posts: 665
Joined: Tue Feb 27, 2007 9:30 am

Re: CD 5-Year Report Card

Post by mur44 » Tue Nov 03, 2015 10:25 am

Kevin M,

The report is simply outstanding. It answered
a lot of my questions. I will keep three reports together.

I am wondering how VICSX (Vanguard Interm-Tm Corp Bd Idx Admiral)
compares with VFIDX (Vanguard Interm-Term Invmt-Grade Inv)
in terms of credit risk. Is it a score of '2' for both?
If it is '2' for both, is there a reason not to go for higher
yield for VICSX?

Thanks a lot.

User avatar
Kevin M
Posts: 10002
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: CD 5-Year Report Card

Post by Kevin M » Tue Nov 03, 2015 11:53 am

mur44 wrote:Kevin M,

The report is simply outstanding. It answered
a lot of my questions. I will keep three reports together.

I am wondering how VICSX (Vanguard Interm-Tm Corp Bd Idx Admiral)
compares with VFIDX (Vanguard Interm-Term Invmt-Grade Inv)
in terms of credit risk. Is it a score of '2' for both?
If it is '2' for both, is there a reason not to go for higher
yield for VICSX?

Thanks a lot.
Glad you found the blog posts useful!

First a minor correction: VFIDX is admiral shares, not investor shares. Regardless, the credit risk will be the same.

To compare credit risk of funds you can use sources such as Morningstar and Vanguard itself. However, these don't necessarily show enough resolution to see the fine differences; e.g., M* shows both Treasury funds and Total Bond Market Index fund as AA, and Vanguard shows them both as highest quality (lowest credit risk), yet TBM is riskier, since it holds about 35% in corporate bonds. You also can view the distribution of fund holdings by credit rating on the Vanguard site; you may be able to eyeball the differences, but you may have to put it in a spreadsheet or something for a finer view.

Without even looking, I'd say that VICSX is riskier, since it is a pure corporate bond fund, whereas the investment-grade bond funds typically hold some Treasuries; also, I know that the yield on the corporate fund is significantly higher, so unless duration is significantly longer, the higher yield must be due to the credit risk. The last comment answers one of your questions: if bond fund yield is higher, risk is higher. Now I'll look ...

First, go to All Vanguard mutual funds by asset class | Vanguard, click Bond in the filter on the left, click the middle cell in the 9-cell grid (intermediate term, medium quality), and we're shown the two funds of interest (only). This shows investor shares for int-term inv-grade (VFICX), but we can show admiral shares by clicking $50,000 and up in the Fund minimum filter.

Yield on VICSX is 3.39% and yield on VFIDX is 2.68%, so we know the former probably is riskier. To see durations, click Attributes & Expense Ratios, then click Bond Attributes. Note that VICSX duration is 6.4 years while it is 5.5 for VFIDX, so duration is at least one reason for the higher risk and higher yield of VICSX.

To view holdings by credit quality, click the fund (I'll start with VICSX), then either scroll to bottom of Overview screen, or click Portfolio & Management to see more details in addition to credit quality distribution. I'll open a Google Sheet and copy/paste the Distribution by credit quality into it, for easy comparison. Now repeat for VFIDX.

Right off the bat I see that VFIDX has 18.5% in US government and AAA, while VICSX has only 1.7% in these top quality categories. I also see that VFIDX has 80.2% in A (actually A3) or above, while VICSX has only 47.9% in A or above (much less resolution of credit ratings is shown for VICSX, but it's good enough). So we can see that VFIDX is indeed higher average credit quality than VCISX.

Here is a link to the spreadsheet in which I did the quick analysis. You can make a copy if you want to edit it.

I thought about including the corporate funds in the blog post comparisons; actually they would be better for medium quality since they are pure corporate, and I used pure Treasury funds for the high quality (low credit risk) category. But I actually own the investment grade funds (int-term and long-term, no short-term), and they've been around longer, so I used them instead. I just included TBM, since it's such a popular fund with Bogleheads.

Checking M*, we see that they rate VFIDX A, and VCISX BBB, so this agrees with the above analysis (at least in rank order). I rated VFIDX a 2 using my credit risk scale (Upper medium grade), so VCISX would be a 3 (Lower medium grade).

Hope that helps!

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

EricBackus
Posts: 18
Joined: Wed Jul 15, 2015 10:30 am

Re: CD 5-Year Report Card

Post by EricBackus » Tue Nov 03, 2015 2:23 pm

Thanks for your work, it is very interesting!

I'm curious about the "return/risk" metric that you use to compare the CD with bonds. Is there some theory for why this is a good metric? Suppose instead you used "return/(risk^2)", or "return*2 - risk". Not that I'm saying those would be useful metrics, but is there some explanation or theory for why return/risk is better?

Similarly, I wonder about adding term risk and credit risk. Do you have any reasoning for why those are the proper numerical values for credit risk? Couldn't you just as easily have used values that were half of what you used? Or twice as big? How do you decide the right scale so that you can add these risks together in a meaningful way?

Sorry to be asking so many questions...

Eric

User avatar
raven15
Posts: 331
Joined: Sun Nov 30, 2014 8:01 pm

Re: CD 5-Year Report Card

Post by raven15 » Tue Nov 03, 2015 3:56 pm

Kevin, is there any particular reason to use direct CD's instead of brokered CD's?
It's Time. Adding Interest.

boglerdude
Posts: 617
Joined: Fri Jul 10, 2015 1:28 am
Contact:

Re: CD 5-Year Report Card

Post by boglerdude » Tue Nov 03, 2015 10:00 pm

What are some real world examples of bond fund risk.

Eg if you put $10k in VICSX in 2006, how much would you lose from defaults over the next few years?

thx1138
Posts: 774
Joined: Fri Jul 12, 2013 2:14 pm

Re: CD 5-Year Report Card

Post by thx1138 » Tue Nov 03, 2015 11:06 pm

raven15 wrote:Kevin, is there any particular reason to use direct CD's instead of brokered CD's?
Read here:

https://www.bogleheads.org/wiki/Certifi ... okered_CDs

Short answer, brokered CDs are typically less attractive both by interest rate and the risk associated with withdrawing before maturity.

jalbert
Posts: 3585
Joined: Fri Apr 10, 2015 12:29 am

Re: CD 5-Year Report Card

Post by jalbert » Wed Nov 04, 2015 12:13 am

Nice work. I would add that bond funds roll their portfolio so they usually get a higher return than holding a single bond to maturity.

Thus, for the 5 years ending 9/30/2015 the Vanguard intermediate treasury bond fund (vfiux) had an annualized return of 2.73%, and the Vanguard GNMA fund (vfijx) had an annualized return of 3.29%.

-jalbert

User avatar
Kevin M
Posts: 10002
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: CD 5-Year Report Card

Post by Kevin M » Wed Nov 04, 2015 2:30 am

raven15 wrote:Kevin, is there any particular reason to use direct CD's instead of brokered CD's?
I need to do a blog post on it. I actually answered an email about it recently, which could be the draft of a blog post, so I'll copy that below. Short answer is that the early withdrawal option of a good direct CD significantly lowers the term risk.

A brokered CD is essentially like an individual Treasury security, so you could substitute 5-year brokered CD for 5-year Treasury in the blog posts, except that you probably would have gotten a better yield on the brokered CD (as is the case now), and Treasuries are more liquid. I actually discussed this some in part 2 of the blog post series.

Here's the answer from the email I mentioned.

------------------------------ Direct CDs vs. Brokered CDs ------------------------------

Rates

Often you can find slightly better yields on direct CDs (purchased directly from bank or credit union), but currently brokered CDs are competitive; 2.25% APY is a good rate.

The 2.25% yield is excellent compared to a 5-year Treasury, with current YTM of about 1.5%. So either direct or brokered is a great deal from this perspective.

One thing you can do with brokered CDs is extend maturity to say 10 years, and find yields of about 3%. Of course this is increasing your interest-rate risk even more, so I just stick with bond funds for higher yields, higher risk, and better liquidity (with about 25% of my fixed income); more on risk in next section.

Risk

This is the main aspect in which direct CDs are the clear winner. Although both types of CDs have no credit risk (assuming within FDIC limits), the brokered CD has much more interest-rate risk (aka, term risk).

A Synchrony Bank 5-year CD with APY 2.25% has an early withdrawal penalty (EWP) of 180 days of interest. So say rates were to increase quickly by 1%, you could pay the penalty of about 1.13% to reinvest at the higher rate.

By contrast, if rates were to increase quickly by 1%, the value of a brokered CD would fall about 4.6% in value, so about four times as much as the EWP of the direct CD. There would be no benefit to selling the CD to reinvest at the higher rate, since you would come out about the same by holding the current CD--actually you'd be better off just holding because you'd lose another 1%-2% to the broker on the sale.

The longer you hold the CD before the rate increase, the less the impact, so the direct CD is mainly a hedge against rates rising significantly within the first few years.

However, another possibility is that a much better rate is offered for a special CD deal, in which case you can pay the 1.13% penalty to invest at the special rate. I actually did exactly this a couple of years ago when PenFed offered a 5-year CD at 3%; I did early withdrawals from three CDs that had yields closer to 2% to buy the 3% CD. With an EWP of 1.13%, you earn back the penalty in slightly over one year, and then are earning an extra 0.75% to 1% over the remaining four years.

I always must mention the caveat that most CD terms allow the bank or CU to deny an early withdrawal, or to change the terms on existing CDs. This has only happened twice in recent years (a few years ago), but some people are concerned it would be more common if rate increased much. On the other hand, 5-year Treasury rates increased by more than 1% in 2013, and there were no new reports of this happening.

Convenience

In taxable accounts, I find buying direct CDs to be just as convenient as buying brokered CDs, except maybe for the additional paperwork to open an account at a credit union. Banks are easy, usually all online. Once account is opened, it usually only takes a few mouse clicks to buy a CD.

In IRAs it's a bit more work, and takes a total elapsed time of about 2-3 weeks to complete, but I still find it well worth it. Once account is opened, I fill out a 1-page IRA transfer form, sign it, and mail it to the new custodian. They take it from there, mailing it to the broker (have done multiple from Vanguard and Fidelity) or other bank or CU, then a check is mailed from current to new custodian, and it's done.

Granted, it is easier to do just the online work to buy a brokered CD in an IRA, so you have to trade off convenience against the extra risk.

Regarding ladders, because the early withdrawal option makes the term risk of a direct CD as low or even less than a shorter-term brokered CD, I just skip the laddering, and always to with at least 5-year maturity to get the higher rates. Also, you usually come out better doing an early withdrawal from a good direct 5-year CD than what you'd earn on a 1-year, 2-year, 3-year or 4-year CD.

FDIC Insurance

If more than $250K in IRA CDs, you're right that this is easier to manage with brokered CDs, as long as you pay attention and track them carefully.

Although there are easy ways to bump this limit to $1.25M or higher in a taxable account, there's no way to do it in an IRA account. So I just live with using multiple banks and credit unions.

Maintenance and Estate

The maintenance is really nothing more than adding a row to my portfolio spreadsheet once the CD is open, then dealing with it five years later when the CD matures. So far, it is well worth it to me to do this to get the occasional much better yield, and more importantly to keep my term risk low while earning the same or higher yield--not only for me but also for a couple of family members and a couple of friends I help.

I have instructions in a Google doc shared with my successor trustees, informing them about the spreadsheet (which I also could share with them), and how to deal with the various classes of assets (trust, IRA, POD). So yes, they would have to contact more institutions if I were to die before simplifying, but hey, they'd be getting well paid to do so (since they also are beneficiaries).

Bottom Line

So, although I've purchased brokered CDs in the past, and actually still have one with a 5% coupon, I have bought only direct CDs over the last five years for the reasons outlined above.

I wouldn't be surprised if I simplify as I get older, or perhaps if rates increased to more historically normal levels, but I'm not ready to do that yet.

Let me know if this helps, and if you have any more questions. I think I've just written the first draft of a blog post on this topic!

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

User avatar
Kevin M
Posts: 10002
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

CD 5-Year Report Card: risk metrics

Post by Kevin M » Wed Nov 04, 2015 2:47 am

EricBackus wrote:Thanks for your work, it is very interesting!
You're welcome; glad you enjoyed it.
I'm curious about the "return/risk" metric that you use to compare the CD with bonds. Is there some theory for why this is a good metric? Suppose instead you used "return/(risk^2)", or "return*2 - risk". Not that I'm saying those would be useful metrics, but is there some explanation or theory for why return/risk is better?
As I mentioned in the blog post, it's basically similar to Sharpe ratio, which is a widely used risk-adjusted return measure--perhaps the most widely used. I provided a link to the Investopedia definition of Sharpe ratio, but here it is again: Sharpe Ratio Definition | Investopedia.

The risk-adjusted return metric I used doesn't subtract a risk-free rate from the asset return in the numerator, and uses a different risk measure in the denominator, but the concept is similar.

I'm not an academic, so perhaps this could be shot full of holes from an academic perspective, and perhaps another alternative would be better, but it seems to me to be a pretty reasonable metric.
Similarly, I wonder about adding term risk and credit risk. Do you have any reasoning for why those are the proper numerical values for credit risk? Couldn't you just as easily have used values that were half of what you used? Or twice as big? How do you decide the right scale so that you can add these risks together in a meaningful way?
I thought some about this, and it just seemed like a relatively simple but somewhat meaningful way to do it. Again, no deep academic justification for it, and would be glad to entertain an alternative that had a more rigorous justification.

As I mentioned in the post, I did take a look at standard deviations of historical returns (perhaps the most common investment risk measure, flawed though it may be) for a few of the funds, and the total risk values I came up with seemed at least roughly in line with them--certainly in terms of rank order (as I mentioned).
Sorry to be asking so many questions...
They're very good questions. Sorry I don't have better answers, but this is where forum members have the opportunity to recommend better analytical approaches.

Kevin
Last edited by Kevin M on Wed Nov 04, 2015 2:59 pm, edited 1 time in total.
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

small_index
Posts: 631
Joined: Sun Aug 16, 2015 2:47 am

Re: CD 5-Year Report Card

Post by small_index » Wed Nov 04, 2015 3:20 am

...
Last edited by small_index on Sat Dec 12, 2015 1:23 am, edited 1 time in total.

User avatar
Kevin M
Posts: 10002
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: CD 5-Year Report Card

Post by Kevin M » Wed Nov 04, 2015 3:24 am

boglerdude wrote:What are some real world examples of bond fund risk.

Eg if you put $10k in VICSX in 2006, how much would you lose from defaults over the next few years?
For real world examples, you can look to periods where different types of bonds didn't do so well. I mentioned 2013, but that was fairly minor. Late 2008 is a better example of where corporate bonds performed very poorly, and the poor performance was correlated quite well with credit risk, and for bonds that had credit risk, also term risk. You can see this by using Portfolio Visualizer to look at the drawdowns for the VG short-term, int-term, and long-term investment-grade bond funds in Oct and Nov of 2008: Backtest Portfolio Asset Allocation. Drawdowns for this 2-month period were:
  • STIG: -6.5%
  • ITIG: -12.0%
  • LTIG:-14.4%
Of course they recovered, but we didn't know at the time that that would be the case--certainly not so quickly. I actually was buying corporate and investment grade bond funds in late 2008 (as well as stocks), but that's not to say that I wasn't experiencing some anxiety, as were many folks then.

Don't know what the actual default rates were in 2008, but we know that companies and financial institutions were failing at a much higher rate than in good times, and there was significant concern (to put it mildly) that the global financial system was at risk of collapse. It's easy to dismiss this with hindsight bias, but it isn't so easy when you're actually living through it, and can't see into the future. Go back to the 1930s for a really bleak "real-world" view; recency bias tends to result in us forgetting about really bad times that have happened in the more distant past.

Basically, to see the real-world risks, just look at performance of bond funds with various levels of the two different types of risks when those risks are showing up.

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

User avatar
Kevin M
Posts: 10002
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: CD 5-Year Report Card

Post by Kevin M » Wed Nov 04, 2015 3:53 am

jalbert wrote:Nice work.
Thanks!
I would add that bond funds roll their portfolio so they usually get a higher return than holding a single bond to maturity.
In the blog post series I discussed how a bond fund is similar to a rolling bond ladder. A ladder may get a higher return when term risk is rewarded, but it depends on a number of factors, like when term risk shows up if it shows up, change in yield curve, etc.

Yield curve change plays a huge roll in this. Assuming a static yield curve, a 1-5 year bond ladder actually would earn about the same return as a 5-year bond held to maturity. If there's a large parallel shift up in the yield curve or if the yield curve flattens significantly due to increasing shorter-term rates toward the end of the 5-year period, the individual bond will do better. Conversely, if yields drop toward the end of the 5-year period, or the yield curve shape changes in the right way, the ladder will do better.
Thus, for the 5 years ending 9/30/2015 the Vanguard intermediate treasury bond fund (vfiux) had an annualized return of 2.73%, and the Vanguard GNMA fund (vfijx) had an annualized return of 3.29%.
Not sure how this illustrates single bond compared to bond fund, since it's two bond funds. Seems to me this just illustrates that whatever additional risks were taken in the GNMA fund were rewarded, just like term risk and credit risk both generally were rewarded over this period.

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

User avatar
Kevin M
Posts: 10002
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: CD 5-Year Report Card

Post by Kevin M » Wed Nov 04, 2015 4:43 am

small_index wrote:One potential correction: quoting the part 3 report, you "assigned a value of 0 to all federally-backed securities (Treasury note and CD)", but then the 99.7% treasuries in Vanguard Long-Term Treasury Fund gets assigned a credit risk of 2.
Yes, good catch. I'll fix it in the blog post, but it doesn't change the results much, since the very high term risk dominates the overall risk, which also is reflected in the standard deviation of returns.
I think it's better to expose the risks you're describing. For example, you created a number called "credit risk", and assign 0-5. But you don't explain why that's meaningful. You could explain why you didn't use the chance of default, or why you didn't apply some multiplier to the difference between corporate and treasury yields.
Yeah, I had to draw the line somewhere in terms of how analytical to get, and how much to explain. I do recall reading that the credit-rating scale is not linear, but as I explained, I just used the Wikipedia article on bond credit ratings to assign the values (so pretty much linear as you move down the ratings).

I did point out that the gap between A and B (investment grade and high yield), was larger, for example than that between AA and A, and spread those numeric values out accordingly.

A more analytical approach would be to have the numeric values better reflect historical default rates, as you imply. Just not enough time or energy to do that for this already humongous series, but maybe a future improvement.
(article) "To quantify term risk for the funds, I used the duration values from Vanguard's web site."
That would happen if rates rose 1% at the end of year 5, yet that seems unlikely.

Duration doesn't say anything about when rates change. It does pretty much relate to an instantaneous, or at least very quick, change. So I don't think I said anywhere that the value of the fund would increase or decrease by a particular amount over the 5-year holding period, but I can see how that might not be clear.

I won't quote your examples related to this point, but they're basically describing total return (price change plus income). The point of the discussion in the blog post is just to explain the concept of duration, and note that this is the measure of term risk being used. Duration is probably the most standard measure of term risk there is, but I really can't go into much more detail about it in a blog post that's already too long.
For the CD, you want to also model how an investor reacts to mitigate loss. I really like how you point out investors can break a CD to capture a higher rate, and you should model that here. If rates rise 1% on year 2.5, an investor can spend 2 months of interest to break the CD and invest in a new CD.
Exactly, but again, the blog posts (parts 2 and 3 anyway) already are quite long for blog posts, so I had to draw the line in terms of going into more detail about this. I was thinking more in terms of just getting some basic theory across, without going into a lot of specific examples. This could be considered the foundation for the kinds of examples you're sharing.
Essentially, term risk hurts bonds of intermediate or long term, but actually causes a gain for the 5 year CD. I think the table should reflect that: +2% for CD, +0% ST bond, -2% IT bond, and -13.6% LT bond. I think those numbers are more meaningful than copying the duration, since it reflects a 1% rise in interest rates at the mid-point.
Appreciate the input, and understand how you think some concrete examples would be more meaningful than more theoretical numbers. However, again, duration doesn't say anything about when a rate change occurs, but I understand the way you're thinking about it--from a more concrete, example-based, total-return point of view.

Rather than saying term risk causes a gain for the CD, I'd just say that the much lower term risk has the potential to hurt it much less (if rates increase much).

I'm not checking your numbers, but you are making the right point from the point of view you're using to present it. I've probably written thousands of BH posts about the benefits of direct CDs, and many of them take an approach more similar to the one you favor.

For example, here's a BH post from early 2012 in which I used a more scenario-based approach to explore and illustrate the benefits of CDs, even if early withdrawal aren't allowed (which would be the case for a brokered CD if it offered the same yield): (1) CD vs. Bond Fund: what if no early withdrawal?. Granted, this is not the same point you're making, and that I usually make when discussing the benefits of direct CDs, but it's an example of the more scenario-based approach you favor.

I definitely was in a more theoretical frame of mind in writing this blog post series. Hopefully it will help lay a useful theoretical foundation to help some folks better understand the risk/return trade-offs of fixed income. But you're right--more examples are required to illustrate the topics more clearly. Maybe a subject for future blog posts, if not already covered in past ones. Speaking of which ...

I just did a Google search of my blog site for examples of more scenario-based explanations. Here's one I wrote about five years ago, shortly after I bought the CD featured in the blog series: Ally Bank 5 Year CD. Note that I only used the word "duration" once in passing (and in parentheses!), and that the explanations are much more examples and scenario-based. So it's not like I haven't explained it in these terms before ;-)

Thanks,

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

User avatar
raven15
Posts: 331
Joined: Sun Nov 30, 2014 8:01 pm

Re: CD 5-Year Report Card

Post by raven15 » Wed Nov 04, 2015 3:41 pm

Thanks Kevin. I previously searched the website, but the first three pages of search results didn't turn up anything nearly as useful as your reply.
It's Time. Adding Interest.

User avatar
Kevin M
Posts: 10002
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: CD 5-Year Report Card

Post by Kevin M » Wed Nov 04, 2015 3:51 pm

I thought it would be worthile to expand some on my answers to Eric's questions about the risk metrics I used in this reply: CD 5-Year Report Card: risk metrics.

I mentioned Sharpe ratio as the model on which I based my risk-adjusted return metric (return / risk). I actually wrote a Bogleheads Wiki article on risk and return based on the type of presentations you find in investment textbooks, and in that article I discussed Sharpe ratio in this section: Risk and return: Portfolio of risky assets combined with a risk-free asset: expected return in terms of risk.

So quick summary is that if you plot return vs. risk, with risk measured as standard deviation (SD), the Capital Allocation Line (CAL) is a line drawn from the y-intercept (risk- free rate/return) to the risky asset or portfolio. You can create a portfolio of desired return/risk by combining the risk-free asset with the risky asset, using any combination along the CAL.

Sharpe ratio is the slope of the CAL, so (Rr - Rrf)/(SDr - SDrf) = (Rr - Rrf)/(SDr - 0) = (Rr - Rrf)/SDr.

If the risk-free rate, Rrf, is 0% as it pretty much has been over the last five years (using short-term Treasury or money market fund as nominal risk-free asset), Sharpe ratio simplifies to Rr/SDr, so return divided by risk. That's the basis for using return divided by risk for the risk-adjusted return metric, as opposed to using some other formulation, although there are other risk-adjusted return metrics, such as Sortino ratio.

I just substituted my overall risk measure for SD in the return/risk risk-adjusted return metric.

I did some analysis this morning to look at how using my overall risk measure compares to using SD to analyze risk-adjusted return. I'll present the results below, but first I want to emphasize that a main purpose of the blog post series was to help people understand fixed-income risk-adjusted returns in the context of the two main types of fixed-income risk: credit and term. So I thought it would be better to create overall risk and risk-adjusted return metrics incorporating these two types of risks, even though they may not be perfect. The main conclusion was that the rank order of returns was pretty well correlated with the overall risk metric, which is what we expect to see with risk-adjusted returns.

Also keep in mind that either type of fixed-income risk may not have dominated the volatility of the returns over this particular holding period. For example, we saw credit risk show up much more in late 2008 than we did over the last five years, so credit risk would have had more impact on the SD of returns for a period including late 2008 than it did over the last five years. Similarly, term risk will have more impact when rates are more volatile. So SD of historical returns doesn't really help us decompose the risk into the two components--at least not without deeper analysis.

Having said that, below are two charts showing return vs. risk using the two different risk measures. I added the return (0.01%) and risk (0%) of the Treasury market fund as the risk-free asset. Note that the trend line is not a CAL, since it doesn't pass through the risk-free rate at the Y-axis, nor does it relate to any one risky asset, but I thought I'd include it anyway.

First, the chart of return vs. risk using my overall risk measure:

Image

Next, the chart using SD as the risk measure (I just fudged some SD values that I think are kind of reasonable for the CD and 5-year Treasury note):
Image

Note that the results from the two approaches aren't that crazily different, and that the trendline fit actually is a bit better using the overall risk measure. So I think that even though the blog post was more focused on presenting a conceptual understanding with some quantification to help, the results don't seem to be bad when comparing to a more standard type of risk-adjusted return analysis.

Hope this helps,

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

herpfinance
Posts: 208
Joined: Tue Nov 18, 2014 4:52 pm
Location: Denmark

Re: CD 5-Year Report Card

Post by herpfinance » Wed Nov 04, 2015 3:52 pm

Thank you very much, Kevin.

Your report has definitely improved my ability to evaluate various fixed income options.

I read all three parts with great interest :sharebeer
"The intelligent investor is a realist who sells to optimists and buys from pessimists" - Benjamin Graham

User avatar
Kevin M
Posts: 10002
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: CD 5-Year Report Card

Post by Kevin M » Wed Nov 04, 2015 3:58 pm

raven15 wrote:Thanks Kevin. I previously searched the website, but the first three pages of search results didn't turn up anything nearly as useful as your reply.
Glad you found it useful! (This was with respect to direct vs. brokered CDs).

I really do plan to use the email reply I posted up-thread as a first draft of a blog post, hopefully to be done in the not-too-distant future. It would be nice to have something to link to to answer this recurring question from my point of view, which I don't think the BH Wiki article really captures. Don't get me wrong--I think the BH Wiki is great, but I've learned that writing good Wiki articles is really hard, especially if you need to get consensus on points of view that aren't backed up by authoritative references. So I just publish blog posts (a bit) and post in the forum (a lot) instead these days.

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

User avatar
Kevin M
Posts: 10002
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: CD 5-Year Report Card

Post by Kevin M » Wed Nov 04, 2015 4:05 pm

herpfinance wrote:Thank you very much, Kevin.

Your report has definitely improved my ability to evaluate various fixed income options.

I read all three parts with great interest :sharebeer
Yay! :D You're welcome, and glad you found it useful. That's what I'm going for!

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

small_index
Posts: 631
Joined: Sun Aug 16, 2015 2:47 am

Re: CD 5-Year Report Card

Post by small_index » Wed Nov 04, 2015 4:07 pm

...
Last edited by small_index on Sat Dec 12, 2015 1:23 am, edited 1 time in total.

jalbert
Posts: 3585
Joined: Fri Apr 10, 2015 12:29 am

Re: CD 5-Year Report Card

Post by jalbert » Wed Nov 04, 2015 4:58 pm

CDs have alot going for them right now, but they are not risk free. If rates are lower in 5 years, you will be facing a second turnover at lower rates after then having realized a 10-yr underperformance in comparison to vfiux (intermediate treasury fund). You have to look at it from the point of view of risk of increased cost of covering liabilities instead of risk of loss of principal to analyze this risk properly.

-jalbert

mammonman
Posts: 70
Joined: Mon Feb 11, 2008 6:41 pm

Re: CD 5-Year Report Card

Post by mammonman » Wed Nov 04, 2015 5:28 pm

Kevin,

Wondering if you could incorporate your thoughts on comparing risk adjusted direct CD returns with high grade muni funds taking into account after tax return. Specifically, comparing them to the Vanguard Intermediate term tax-exempt fund VWIUX.

Thanks - Really enjoy your analysis

User avatar
Kevin M
Posts: 10002
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: CD 5-Year Report Card

Post by Kevin M » Wed Nov 04, 2015 5:49 pm

small_index wrote:Thanks for the links to other articles. On a side note, anyone looking at Ally's 5 year CD today faces 150 day EWP, rather than the 60 day EWP from 5 years ago (now only 2 yr CD at Ally have 60 day EWP). Looking forward to see how Ally changes their CD rates on Saturday (Nov 7).
You're welcome!

Yeah, I stopped buying Ally CDs a few years ago, since there always were higher yields with only slightly higher EWPs (e.g., 6 months instead of 5). I did let one small IRA CD renew, but will transfer proceeds from most renewals to other banks or CUs unless they bump their rates a bit.

They do give a 0.05% "loyalty reward" bump upon renewal, but still are about 20 basis points short of competitive. However, if I were too lazy to do the IRA transfers, or just didn't care about the extra few dollars, I'd be OK letting them renew. Still quite a bit better than a 5-year Treasury at 1.64%, although the gap has been shrinking in the last couple of weeks (5-year Treasury low of 1.29% on Oct 14).
Duration shows the magnitude of gain/loss when interest rates change. But in the risk/return table, you mix it with an actual return from the past 5 years. How can duration remain theoretical, while being mixed with a real yield?
Duration is just a measure of risk, in this case, term risk. This is similar to using standard deviation (SD) as a measure of risk, and note that the results I showed up-thread using SD come out pretty darn similar to my own metric of total risk (credit risk + duration).

It is standard practice to use a theoretical risk measure for estimating risk-adjusted returns, and to calculate risk-adjusted returns using annualized historical returns divided by the risk measure, such as Sharpe ratio uses SD in the denominator. Read a book or article by Larry Swedroe and you'll find it filled with data using the same approach.

Now, I do have a bit of a beef with using annualized SD in evaluating the risk of returns for holding periods longer than one year, even though it is done all the time. The relevant measure is SD of the returns over the holding period (and the relevant returns are the returns over the holding period, not annualized), but since we don't have enough longer-term holding periods to have any meaning statistically, it is common practice to ignore this issue and just use monthly or annual SD and annualized returns.

All of your examples are just digging you deeper into a hole of trying to apply a theoretical risk measure in a way it's not intended to be used to calculate something related to total returns. If what I'm doing is so off-base, the results wouldn't be so consistent with the risk-adjusted return relationships using the more standard metric of SD.

Once again, the one of the main points of the blog post series is to help people think clearly in terms of fixed-income return vs. risk, using the two standard risk dimensions for fixed income. Attaching some numerical values to the two risk dimensions just helps quantify it, hopefully making it more clear, but apparently not in your case :-( .

If you understand the concept of risk adjusted return, e.g., as expressed using Sharpe ratio, the graphs I posted up-thread should help clarify all of this. If they don't, then we should focus on how the graphs do or don't illustrate the standard relationship of risk and return in investments, which is the thing that's most important to understand.

A big problem I see quite often is that people tend to say, "my bond fund had a higher return than your CDs, so it's better", but they don't consider that the higher return was achieved by taking more risk. This is a misunderstanding I'm trying to correct.

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

User avatar
Kevin M
Posts: 10002
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: CD 5-Year Report Card

Post by Kevin M » Wed Nov 04, 2015 6:00 pm

jalbert wrote:CDs have alot going for them right now, but they are not risk free. If rates are lower in 5 years, you will be facing a second turnover at lower rates after then having realized a 10-yr underperformance in comparison to vfiux (intermediate treasury fund). You have to look at it from the point of view of risk of increased cost of covering liabilities instead of risk of loss of principal to analyze this risk properly.

-jalbert
Nothing is really risk free, unless you clearly specify the unit of account (e.g., dollars, purchasing power, etc.) and holding period. A 5-year Treasury or CD is pretty close to risk free in nominal terms for a 5-year holding period. A 5-year TIPS is pretty close to risk free in real terms for a 5-year holding period. So depending on whether your liability is nominal or real, the risk-free investment is a nominal or real fixed income investment with a maturity date that matches the liability.

A bond fund isn't risk-free for matching liabilities in either nominal or real terms, but you can approximate a liability matching ladder with a combination of bond funds of different durations and cash or other very short-term maturity fixed income.

Comparing the 10-year return of a 5-year CD or individual Treasury to the int-term Treasury fund is not really an appropriate comparison, since the fund holds bonds in the 3-10 year range, and is therefore quite a bit riskier than a 5-year CD.. A more appropriate comparison would be of a 10-year Treasury to the fund. Under static yield-curve assumptions, they will return about the same. No one knows what the yield curve will do, so you could get bit either by price risk or reinvestment risk, both of which are subsets of term risk (interest-rate risk).

So no, the 5-year CD isn't risk free, but the risk is much lower than for int-term Treasury fund. Risk has an upside (or you shouldn't take it), so it could pay off over the next 10 years, and the fund could do better--just like both credit and term risk have paid off over the last five years, and the higher risk bond funds generally did better.

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

User avatar
Kevin M
Posts: 10002
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: CD 5-Year Report Card

Post by Kevin M » Wed Nov 04, 2015 6:12 pm

mammonman wrote:Kevin,

Wondering if you could incorporate your thoughts on comparing risk adjusted direct CD returns with high grade muni funds taking into account after tax return. Specifically, comparing them to the Vanguard Intermediate term tax-exempt fund VWIUX.

Thanks - Really enjoy your analysis
Absolutely! I actually included returns from a few tax-exempt bond funds in the spreadsheet I created from the Vanguard website, but decided it would be too confusing to mix taxable and tax-exempt returns in this post series.

Suffice it to say that tax-exempt bond funds have done quite well over the last five years, especially the CA funds, which I own (I also owned the national int-term and long-term funds, but have recently sold my last shares to buy more CDs, but still own a few shares of the high-yield tax-exempt bond fund, although those will go soon too, since I'm now under the admiral shares limit). These funds have outperformed their taxable peers even on a pre-tax basis, so after tax, quite stellar.

But the same principle applies: there is no free lunch when it comes to marketable securities that institutional investors can buy and sell to arbitrage away any inefficiencies (unlike CDs). The higher returns have come at the cost of taking more risk. If you remember the Meredith Whitney scare a few years ago, that's when the risk showed up.

So tax-exempt bond funds comprise part of the 25% of fixed income I hold in bond funds, but I also hold CDs in taxable accounts, just as institutional investors hold Treasuries in taxable accounts.

One guideline is that tax-exempt bond funds are priced to be competitive at a federal marginal tax rate of 25% (or so states forum member grabiner, who seems to be quite knowledgeable in these things). So I'd compare them to CDs and other bond funds in about the same way, but calculating the taxable-equivalent yield using a 25% tax rate. If the taxable-equivalent yield of the tax-exempt bond fund at 25% marginal tax rate is higher than the yield of a taxable bond fund you're comparing to, then the market probably is telling you that it is riskier.

If in a higher tax bracket, the return/risk could be somewhat in your favor with tax-exempt bonds.

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

castlemodesto
Posts: 144
Joined: Sun Sep 15, 2013 12:31 pm

Re: CD 5-Year Report Card

Post by castlemodesto » Wed Nov 04, 2015 8:44 pm

Thanks Kevin for those excellent papers on CDs and Bonds. Very thorough.
I have access to a Stable Value fund paying 3.5%. It is with a very highly rated company ( nationwide) so there is a very low risk of default. Also there are no liquidity restriction ( which can be an issue sometimes for SValue funds) .
I am assuming this would put this as having an an even better risk/return than the CD example you used.
The only reason I can see bothering with CDs in this case is the CD return is locked in for the whole period of time where as the SValue could be reduced , (perhaps because of extremely low interest rates) to below the CD rate: which may mean there was lost opportunity to lock in a relatively high returning CDs. For this reason I did purchase some 3% PenFed 7 year CDs (with probably no withdrawal restrictions since over 59.5) to hedge against this.
But given that rising inflation and rising interest rates are more probable, it seems to me that this SValue is preferable to CDs since in a rising interest rate environment the SValue can always be traded in for CDs when CDs have a better (after withdrawal penalty) return.
Have I got that right?
And regarding your suggestion of 25% of fixed income in Intermediate Term Investment grade: would you consider the 4.25% average annual return (but with much higher risk) make it worth doing compared to a 3.5% SValue.
I dont think I have ever seen anyone suggest having 100% of ones fixed income in a SValue fund (apart from the TIAA- Cref Traditional account groupies) so I would be interested in your ( or anyone else )comments. Thanks.

User avatar
Kevin M
Posts: 10002
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: CD 5-Year Report Card

Post by Kevin M » Wed Nov 04, 2015 9:34 pm

castlemodesto wrote:Thanks Kevin for those excellent papers on CDs and Bonds. Very thorough.
Thanks for the kind words!
I have access to a Stable Value fund paying 3.5%. It is with a very highly rated company ( nationwide) so there is a very low risk of default. Also there are no liquidity restriction ( which can be an issue sometimes for SValue funds) .
I am assuming this would put this as having an an even better risk/return than the CD example you used.
Yes, I probably would use that in preference to CDs earning 2.25%, at least for a large portion of fixed income.
The only reason I can see bothering with CDs in this case is the CD return is locked in for the whole period of time where as the SValue could be reduced , (perhaps because of extremely low interest rates) to below the CD rate: which may mean there was lost opportunity to lock in a relatively high returning CDs. For this reason I did purchase some 3% PenFed 7 year CDs (with probably no withdrawal restrictions since over 59.5) to hedge against this.
Makes sense, but even though the SV fund has low credit risk, the credit risk is higher than a federally-backed security. But given that there's no term risk, a little default risk probably is worth earning 3.5%. It certainly is an excellent risk-adjusted return.

If we started getting into serious financial crisis, I might be more cautious about SV funds, since the likelihood of systemic failures would be greater.
But given that rising inflation and rising interest rates are more probable, it seems to me that this SValue is preferable to CDs since in a rising interest rate environment the SValue can always be traded in for CDs when CDs have a better (after withdrawal penalty) return.
Have I got that right?
I always hesitate to make interest-rate predictions. I think the general expectation in recent years has been that rates would increase, yet in general they haven't--certainly not at the short-term end of the curve.

Since 2007, 5-year and 10-year Treasury yields declined until late 2008, shot back up in 2009, then continued the roller coaster ride to the lows of 2012. Then the 5-year rate rose a bit more than 100 basis points, and the 10-year by about 150 basis points, peaking in 2013. Since then, 5-year has been bouncing between about 1.4% and 1.8%, and is about in the middle of that range now; the 10-year dropped from over 3% at the end of 2013 to a low of about 1.7% in early 2015, and now is back at around 2.2%.

Image

I have no idea where they'll go next, but whenever they start hitting historic lows, I tend to shift a bit more from bonds to CDs.

At any rate, since your SV fund yield is higher than the 5-year Treasury has been since mid-2008, higher than the 10-year since early 2011, and significantly higher than a good CD or the TSP G fund, I'd feel pretty good about it.
And regarding your suggestion of 25% of fixed income in Intermediate Term Investment grade: would you consider the 4.25% average annual return (but with much higher risk) make it worth doing compared to a 3.5% SValue.
I'd look more at yield than 5-year return; the blog post series was more to help people who focus too much on past returns understand that those returns haven't been a free lunch.

The SEC yield of VFIDX is 2.79%, and even the distribution yield is only about 3.2%. So you're SV fund clearly is a much better risk-adjusted deal. I think I'd stick with the SV fund.
I dont think I have ever seen anyone suggest having 100% of ones fixed income in a SValue fund (apart from the TIAA- Cref Traditional account groupies) so I would be interested in your ( or anyone else )comments. Thanks.
Well, there is always some default risk, minor though it may be. I have to admit it might make me nervous to have 100% of fixed income in an SV fund, and might keep an eye out for good CD deals to diversify, but of course you can't take advantage of them within the 401k.

You can get yields above 4% in the long-term corporate and investment grade funds. People tend to criticize using long-term bond funds, but they also tend to forget that TBM holds long-term corporate bonds. Adding a touch of them is no worse than holding TBM, and as I often argue for CDs, probably even better, since your SV fund (or my CDs) is generating higher yield than most of the Treasuries in TBM, and with much less term risk (although SV fund has some credit risk). So why not add some bonds with more term risk and maybe even credit risk for a bit of a hedge against continued low rates, or even falling rates, just like the owners of TBM do (although perhaps without realizing it)?

Not necessarily suggesting it, but throwing it out for your consideration.

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

User avatar
Doc
Posts: 8510
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: CD 5-Year Report Card

Post by Doc » Thu Nov 05, 2015 8:36 am

A couple of side thoughts:

People need to be aware that when you are looking at credit ratings of funds you can get different ratings depending on the source. It's the methodology not a data error. M* uses a weighting method that takes into consideration the probability of default not just a "size" weighted average.

I haven't gone through this thread in depth but a brief review seems to be that Kevin is addressing both term and credit risk when looking at the CD vs. Treasury issue. There is also the liquidity risk with CD's. Yes you can get your money back but the CD is not marked to market so in a flight to quality situation the Treasury price may go up significantly while the CD price will go down by at least the cost of the early withdrawal penalty. I have no idea how one would quantify this factor since it concerns the entire portfolio not just the FI. In a recent post Kevin addresses the idea of having sufficient very high quality FI assets to cover this situation. If you do the calculation this is not a very large amount. For me I believe this "bucket" should be in Treasuries not CDs. But if you just turned off the tube during the Lehman crisis and just waited for everything to come back the CD might have been a better choice for you.
A scientist looks for THE answer to a problem, an engineer looks for AN answer and lawyers ONLY have opinions. Investing is not a science.

trueblueky
Posts: 1331
Joined: Tue May 27, 2014 3:50 pm

Re: CD 5-Year Report Card

Post by trueblueky » Thu Nov 05, 2015 12:24 pm

Kevin,
If risk = standard deviation of the value of the underlying instrument, how do you figure SD of an I Bond or G Fund? Their values change, but only in one direction. Do you just pretend the SD is zero?

G Fund gained 2.04% over the past 12 months (thru 31 Oct) -- who knows next year, but it won't be a loss.

I Bonds are paying 1.54% plus the fixed rate for the next six month period, so no short-term variability. If the question is real rather than nominal, no variation ever.

If risk of these is zero, am I right to use their rate as my risk-free rate?

User avatar
Kevin M
Posts: 10002
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: CD 5-Year Report Card

Post by Kevin M » Thu Nov 05, 2015 4:56 pm

Doc wrote: People need to be aware that when you are looking at credit ratings of funds you can get different ratings depending on the source. It's the methodology not a data error. M* uses a weighting method that takes into consideration the probability of default not just a "size" weighted average.
OK, but giving an AA instead of AAA to a fund that owns only Treasuries? As I said in the blog posts, I think this is because of the downgrade in US credit rating by S&P, but I think the market still prices US Treasuries as highest quality.

As I recall, Treasury price went up after downgrade! Checking ... yes, on day of downgrade, August 5, 2011 10-year yield did go up from 2.47% to 2.58% (so price down slightly), but on the following trading day, it resumed its downward trend, dropping to 2.40%, and continued down from there (so price resumed upward trend).

Image

So as far as market concerned, the downgrade seemed to be pretty much of a non-event.
I haven't gone through this thread in depth but a brief review seems to be that Kevin is addressing both term and credit risk when looking at the CD vs. Treasury issue.
Correct--not just for Treasuries, but for bond funds in all top six cells of quality/maturity grid, plus bottom-middle cell (low quality, medium term).
There is also the liquidity risk with CD's. Yes you can get your money back but the CD is not marked to market so in a flight to quality situation the Treasury price may go up significantly while the CD price will go down by at least the cost of the early withdrawal penalty.
Correct, although I'd say that the Treasury may go up when stocks go down a lot. Has not always happened; if you remove 2002 and 2008, 3-month Treasury (i.e., cash) has done as good or better than 10-year Treasury for worst annual stock declines.

I didn't get into this level of detail in this blog post series, but have mentioned this many times in posts here. Also good to keep in mind when considering brokered CDs, since probably will take a hit of 1%-2% if selling before maturity. However, may still end up better off with brokered CD, since the higher yield is likely to more than make up for haircut on sale if you hold for at least 2-3 years, depending on CD/Treasury premium.
In a recent post Kevin addresses the idea of having sufficient very high quality FI assets to cover this situation. If you do the calculation this is not a very large amount. For me I believe this "bucket" should be in Treasuries not CDs. But if you just turned off the tube during the Lehman crisis and just waited for everything to come back the CD might have been a better choice for you.
Agree with all of this. I'd go further and say that if you really want a slice of fixed-income for flight-to-quality hedging, go with long-term Treasuries or fund, since this is what did the best, by far, in late 2008.

But for this to work, you have to be alert for the rebalancing opportunity, as it was quite fleeting in late 2008, and the 10-year Treasury had fully recovered by early June 2009. Someone doing annual rebalancing in July-October would have missed it entirely:

Image

If you missed it, you may have been better of with corporate bonds. I didn't have Treasuries in late 2008, but I had cash, and actually was buying corporate and investment grade bonds, in addition to stocks, during late 2008 and early 2009. Didn't start buying CDs until late 2010, after my bonds had gone up a lot.

Thanks for inputs, Doc!

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

User avatar
Kevin M
Posts: 10002
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Risk free rates: G fund, I bond

Post by Kevin M » Thu Nov 05, 2015 5:27 pm

trueblueky wrote:Kevin,
If risk = standard deviation of the value of the underlying instrument, how do you figure SD of an I Bond or G Fund? Their values change, but only in one direction. Do you just pretend the SD is zero?
Good question.

"Price" actually doesn't change in the conventional sense, any more than the share price of a Vanguard bond fund changes when you reinvest a dividend distribution. G fund "price" is stable, but interest is automatically reinvested, and you see this as an increase in value of your holding. Basically same with I Bonds.

G fund rate changes monthly, so it is a risk-free rate for a one month holding period. Same with a 1-month Treasury. However, even though rate changes monthly, academics use short-term Treasury as the risk-free rate (in nominal terms), since it fluctuates in value so much less than riskier investments.

A 1-year Treasury is nominal risk-free for a 1-year holding period. But if you calculate SD of annual returns, it is not 0%, since return varies with rate each year. Again, academics would treat is as a risk-free (nominal) rate in calculating Sharpe ratios for historical risk-adjusted returns of riskier investments.

Since inflation component of I Bond rate changes every six months, it is risk free in nominal terms for a six month holding period (lately risk-free 0% on newer purchases, but 1.64% for new purchases now). Since real rate on current purchases is locked at 0.1% real for 30 years, it is risk-free rate of 0.1% for 30 years, but since you can get 1.27% on 30-year TIPS, that's a better number for risk-free 30-year real rate.
If risk of these is zero, am I right to use their rate as my risk-free rate?
Hopefully discussion clarifies. As I said in an earlier reply, true risk-free rate depends on unit of account and holding period.

However, I think it would be typical for an institutional investor to use 0% as current risk-free nominal rate, and for a retail investor to use about 1%, since that's what you can earn in a federally-insured savings account.

If you have access to G fund, then sure, that's as much of a risk free rate as a savings account, since neither has any credit risk or term risk (in the sense of quick value change for quick rate change), but as discussed above, technically they are risk-free only for the period the rate is guaranteed.

Interesting to note that good 3-month CD rate is only about 0.5%. Why would anyone hold a 3-month CD unless they expect savings account rates to drop to less than 0.5% over next three months? Go out to six months, and you're back to about 1%, but again, why lock for six months unless you think savings account rates will drop?

By 1-year, CD yield curve is positive again, with good rates at about 1.3%. Even if you break a good 5-year CD after one year, you'll earn 1.17% (Ally Bank) or 1.12% (Synchrony Bank).

Does this help?

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

User avatar
Doc
Posts: 8510
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: CD 5-Year Report Card

Post by Doc » Thu Nov 05, 2015 5:32 pm

Kevin M wrote:Doc wrote:
People need to be aware that when you are looking at credit ratings of funds you can get different ratings depending on the source. It's the methodology not a data error. M* uses a weighting method that takes into consideration the probability of default not just a "size" weighted average.

OK, but giving an AA instead of AAA to a fund that owns only Treasuries? As I said in the blog posts, I think this is because of the downgrade in US credit rating by S&P, but I think the market still prices US Treasuries as highest quality.
Just a few clarifications.

First the AA or AAA for Treasuries is another matter. But the same caveat applies. Don't use bond fund data from different sources for comparison.

The M* methodology uses risk of default not just weighted average so while Vanguard may have a fund of 50% A and 50% C and say the fund is rated B, M* may rate that fund as B minus. Same caveat as above. But this one gets harder to "fix" once we bring in other ratings like CD's.

Some funds make a large use of derivatives to obtain a particular risk profile using short term instruments. M* treats these short derivatives as "cash" but they might actually be an interest rate future or whatever that represents the risk/reward of a three year note. I think this was the issue that made PIMCO stop giving M* data awhile back.

The flight to quality situation is another one hard to quantify. As Kevin says it doesn't always happen and when it does it only occurs perhaps for a few months. But even if you yourself don't rebalance during a market crash there are others that do and their attitude affects the risk/reward profile in normal times. For example I'm willing to take a little less yield with a Treasury compared to a CD and my extremely large trading portfolio therefore affects the Treasury market a smidgen. How much the real traders influence the market because of this effect is unknown.

Anyway Kevin. Good stuff.
A scientist looks for THE answer to a problem, an engineer looks for AN answer and lawyers ONLY have opinions. Investing is not a science.

User avatar
Kevin M
Posts: 10002
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: CD 5-Year Report Card

Post by Kevin M » Thu Nov 05, 2015 10:58 pm

Doc wrote:The M* methodology uses risk of default not just weighted average <snip>
This may relates to the point I made in an earlier reply about the default risk not being linearly related to each step in the credit rating scale. Here's a paper I just found that discusses this exact point, and how it relates to mutual fund average credit quality: www.slcg.com/pdf/workingpapers/Average Credit Quality in Bond Portfolios.pdf.

Looking at the chart on page 3 of the paper, we see that the rate of change for 5-year default rate starts to increase rapidly moving from BBB to lower ratings. Using the scale shown, we can barely see any difference from AAA through A- (six steps lower).
Doc wrote:<snip> ... so while Vanguard may have a fund of 50% A and 50% C and say the fund is rated B, M* may rate that fund as B minus.
Actually, I don't find an average credit rating at all for Vanguard on their site in the overview or portfolio and management pages--just distribution by credit rating. For Treasury funds we just see something like Government 99.4%, NR 0.6%. Government is listed above AAA, with the implication that it's higher quality.

Here is what Vanguard says in the semi-annual report for US government bond funds:
Credit Quality. Credit-quality ratings are measured on a scale that generally ranges from AAA
(highest) to D (lowest). U.S. Treasury, U.S. Agency, and U.S. Agency mortgage-backed securities
appear under “U.S. Government.” For the federal and Treasury funds, credit-quality ratings are
obtained from Moody’s and S&P, and the higher rating for each issue is shown. For the GNMA
Fund, credit-quality ratings are obtained from Barclays and are from Moody’s, Fitch, and S&P. When
ratings from all three agencies are used, the median rating is shown. When ratings from two of the
agencies are used, the lower rating for each issue is shown. “Not Rated” is used to classify
securities for which a rating is not available. Not rated securities include a fund’s investment in
Vanguard Market Liquidity Fund or Vanguard Municipal Cash Management Fund, each of which
invests in high-quality money market instruments and may serve as a cash management vehicle for
the Vanguard funds, trusts, and accounts.
So "U.S. Government" is a rating of its own, and also is listed at the top of the quality ranking in the 9-cell maturity/quality grids in the reports.
Doc wrote:The flight to quality situation is another one hard to quantify. As Kevin says it doesn't always happen and when it does it only occurs perhaps for a few months. But even if you yourself don't rebalance during a market crash there are others that do and their attitude affects the risk/reward profile in normal times. For example I'm willing to take a little less yield with a Treasury compared to a CD and my extremely large trading portfolio therefore affects the Treasury market a smidgen. How much the real traders influence the market because of this effect is unknown.
Isn't this just saying that institutional investors who drive prices in the $500 billion a day Treasury market tend to drive the prices up (and rates down) because they value the high quality of US Treasuries? The same probably would happen to CDs if they were risk-free to institutional investors like they are to retail investors.
Doc wrote:Anyway Kevin. Good stuff.
Thanks! High praise coming from you, Doc.

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

User avatar
Doc
Posts: 8510
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: CD 5-Year Report Card

Post by Doc » Fri Nov 06, 2015 3:48 pm

Kevin M wrote:Actually, I don't find an average credit rating at all for Vanguard on their site in the overview or portfolio and management pages--just distribution by credit rating. For Treasury funds we just see something like Government 99.4%, NR 0.6%. Government is listed above AAA, with the implication that it's higher quality.
I was just using Vg as a generalized "not Morningstar" example. I looked around and various fund companies use different methods and definitions. Vg uses a risk "bar" and Schwab actually uses M*'s style box. All I was trying to point out is that when comparing funds from different providers you need to stick with a common definition of average credit quality when trying to separate credit risk with term risk.
Kevin M wrote:Isn't this just saying that institutional investors who drive prices in the $500 billion a day Treasury market tend to drive the prices up (and rates down) because they value the high quality of US Treasuries? The same probably would happen to CDs if they were risk-free to institutional investors like they are to retail investors.
The problem I have with that argument is that it ignores any liquidity advantage that Treasures have over CD's. You analysis implies that it is the FDIC guarantee limit that is responsible for very large investors to shun CD's. While undoubtedly that is part of the story I'm not sure the nearly "instant" liquidity of Treasuries doesn't' explain part of the difference. Just this week I used our margin account to move $'s in 1 day instead of the 2 it would take from a short bond fund. And I could have gotten it done in minutes if I had used a wire instead of an EFT (ACH?) transfer. How long would it have taken if I have to liquidate a CD? I assume it is longer and there is some cost and therefore some risk however small in having to wait. And that waiting period is not reflected in the credit and term risk in your model. On the other hand for many retail investors especially those who rebalance on a time schedule or who turn off the TV when the stock market crashes it makes no difference at all. :)
A scientist looks for THE answer to a problem, an engineer looks for AN answer and lawyers ONLY have opinions. Investing is not a science.

User avatar
Kevin M
Posts: 10002
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: CD 5-Year Report Card

Post by Kevin M » Sat Nov 07, 2015 7:04 am

I think our dialogue got a little jumbled.
Kevin M wrote:
Doc wrote:But even if you yourself don't rebalance during a market crash there are others that do and their attitude affects the risk/reward profile in normal times. For example I'm willing to take a little less yield with a Treasury compared to a CD and my extremely large trading portfolio therefore affects the Treasury market a smidgen.
Isn't this just saying that institutional investors who drive prices in the $500 billion a day Treasury market tend to drive the prices up (and rates down) because they value the high quality of US Treasuries? <snip>
The main point I think we are both making is that the perception that Treasuries will do well in a flight-to-quality scenario is exactly the reason, or at least a big reason, that Treasury rates are low and prices are high. This is the term risk piece coupled with extremely low default risk; i.e., long-term Treasuries (more term risk) did much better in 2008. The low default risk is key though, since long-term corporates did not do well in 2008. I don't think there is any possible disagreement here.

I then went on to say:
Kevin M wrote:The same probably would happen to CDs if they were risk-free to institutional investors like they are to retail investors.
Maybe I also should have added the criterion that there also would have to be sufficient volume and liquidity for institutional investors to play in the market, and arbitrage away the price premiums over Treasuries. I don't know how the volume of bank borrowing/lending compares to the supply of Treasuries (supply side), but the fact that CDs are not risk-free to institutional investors must negatively impact the demand side.

So does low demand due to higher credit risk affect the liquidity, or does the lack of liquidity affect the demand? In other words, liquidity and credit risk seem to be a bit of a chicken and egg situation, and perhaps it's difficult to sort it out.
Doc wrote:The problem I have with that argument is that it ignores any liquidity advantage that Treasures have over CD's. You analysis implies that it is the FDIC guarantee limit that is responsible for very large investors to shun CD's. While undoubtedly that is part of the story I'm not sure the nearly "instant" liquidity of Treasuries doesn't' explain part of the difference.
And I agree with this, but still am scratching my head over what causes what in terms of liquidity and demand. At any rate, there is no question that Treasuries are much more liquid than CDs. I look at this as just another advantage for the retail investor who has even a modicum of patience. After all, we're not talking about timber land here--the liquidity isn't all that bad.
Doc wrote:Just this week I used our margin account to move $'s in 1 day instead of the 2 it would take from a short bond fund. And I could have gotten it done in minutes if I had used a wire instead of an EFT (ACH?) transfer. How long would it have taken if I have to liquidate a CD? I assume it is longer and there is some cost and therefore some risk however small in having to wait. And that waiting period is not reflected in the credit and term risk in your model. On the other hand for many retail investors especially those who rebalance on a time schedule or who turn off the TV when the stock market crashes it makes no difference at all. :)
And this makes my point. I have personally done an early withdrawal from a direct CD and had the funds available same day or next day, and used those funds to buy a better CD. I could have used them to do a same-day purchase of a Vanguard stock fund from a linked external account.

I just don't find this at all onerous, and if I'm getting paid an extra 100 basis points for this minor inconvenience, it's a pretty sweet deal. Whatever part of the premium is due to this slightly lower liquidity, I'm fine with it.

I'm glad there are investors out there like you (although you are not representative of the investors who are really moving the needle) who are willing to pay top dollar for whatever combination of no credit risk and high liquidity they find attractive, keeping them in that playpen and out of my playpen in which I can get much better deals ;-)

For the extra liquidity I do want or need, I use cash and bond funds. I could use Treasuries, but retail-investor cash actually earns more than short-term Treasuries out to about two year maturity, and I'm not excited enough about flight-to-quality rebalancing opportunities to hold long-term Treasuries, but someday I might be.

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

User avatar
Doc
Posts: 8510
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: CD 5-Year Report Card

Post by Doc » Sat Nov 07, 2015 9:36 am

Kevin M wrote:The main point I think we are both making is that the perception that Treasuries will do well in a flight-to-quality scenario is exactly the reason, or at least a big reason, that Treasury rates are low and prices are high. This is the term risk piece coupled with extremely low default risk; i.e., long-term Treasuries (more term risk) did much better in 2008. The low default risk is key though, since long-term corporates did not do well in 2008. I don't think there is any possible disagreement here.
The disagreement is how you quantify the effect. You are quantifying it by looking at the "long" term difference in term risk and credit risk. I think there is also a very short term risk that is present only occasionally and it is a risk of not gaining more money rather than losing money.

During the Lehman crisis the price of one of the Vanguard Intermediate Term Treasury funds gained some 10% over a very short period of time. If such an occurrence happens once every 10 years and you capture half of it through re-balancing you have on average an additional 50 bps in favor of the Treasury. At the same time because of the EWP on the CD you have a disadvantage of maybe 10 to 20 bps. (You don't even get the 50% capture rate - It's 100%.) I assume since you are discussing a 5 year CD you are comparing it to a 10 year Treasury ladder. With the ladder you have the potential of roll yield that is not available from the CD again because of the EWP. This might amount to another 20 bps. At this point we have negated nearly all of the 100 bps advantage that you assign to the CD.

Now if you don't re-balance religiously during a market crash a lot of this Treasury stuff goes away and Treasuries lose to the CD. But in this case Treasuries are the wrong comparison. If you don't care about now a total bond market fund or even an investment grade fund might suit your need better.

The failure of many of these risk threads is that most retail investors even Bogleheads think of risk only as losing money. But the professionals who are in reality setting the market returns are also thinking of the risk of not gaining money.

Anyway good discussion. Now I need to work on completing my 2015 tax return so that I can plan all my yearend tax gain harvesting.
A scientist looks for THE answer to a problem, an engineer looks for AN answer and lawyers ONLY have opinions. Investing is not a science.

stlutz
Posts: 4664
Joined: Fri Jan 02, 2009 1:08 am

Re: CD 5-Year Report Card

Post by stlutz » Sat Nov 07, 2015 2:24 pm

Just for kicks (it's cold out today), I made a up a bit of a backtest to try and simulate this over a longer period of time. I took the Simba spreadsheet and created a new data series I called "Bank CDs". I calculate each year's return by taking the average of the 5 year Treasury rates for the prior 5 years and then added 1.2% to the result to replicate a 5 year CD ladder. (So, right now I would be saying that new CDs are yielding 3%).

Results for the 1972-2014 period:

40% US stocks/20% Intl stocks/40% Int. Treasury: CAGR 9.79% St. Dev. 10.96
40% US stocks/20% Intl stocks/40% Bank CDs: CAGR 9.75% St. Dev. 11.00

In short, a wash. And you'll see why I picked 1.2% as the CD yield "bonus" over Treasuries--that was what I had to add to make the portfolio as good as one with Treasuries.

Obviously the portfolio with Treasuries did better in times where stock/bond correlations were negative and the CDs did better when correlations were positive.

So, I think what I'm adding to the conversation is that:

a) A treasury fund (equivalent of a 10-year treasury ladder) has more term risk by itself than the CDs do. In a balanced portfolio, the overall risk ended up being a wash.

b) You didn't have to market time as well as Doc suggests to benefit from rebalancing--once per year worked just fine.

Johno
Posts: 1883
Joined: Sat May 24, 2014 4:14 pm

Re: CD 5-Year Report Card

Post by Johno » Sat Nov 07, 2015 4:02 pm

Doc wrote:
Kevin M wrote:The main point I think we are both making is that the perception that Treasuries will do well in a flight-to-quality scenario is exactly the reason, or at least a big reason, that Treasury rates are low and prices are high. This is the term risk piece coupled with extremely low default risk; i.e., long-term Treasuries (more term risk) did much better in 2008. The low default risk is key though, since long-term corporates did not do well in 2008. I don't think there is any possible disagreement here.
The disagreement is how you quantify the effect. You are quantifying it by looking at the "long" term difference in term risk and credit risk. I think there is also a very short term risk that is present only occasionally and it is a risk of not gaining more money rather than losing money.

During the Lehman crisis the price of one of the Vanguard Intermediate Term Treasury funds gained some 10% over a very short period of time. If such an occurrence happens once every 10 years and you capture half of it through re-balancing you have on average an additional 50 bps in favor of the Treasury. At the same time because of the EWP on the CD you have a disadvantage of maybe 10 to 20 bps. (You don't even get the 50% capture rate - It's 100%.) I assume since you are discussing a 5 year CD you are comparing it to a 10 year Treasury ladder. With the ladder you have the potential of roll yield that is not available from the CD again because of the EWP. This might amount to another 20 bps. At this point we have negated nearly all of the 100 bps advantage that you assign to the CD.

Now if you don't re-balance religiously during a market crash a lot of this Treasury stuff goes away and Treasuries lose to the CD.
Yes if you don't re-balance at all, you gain nothing from the liquidity of treasuries. But it's not binary. Even if you rebalance, the advantage (in actual 'money in the bank') of treasuries is only for the portion of them you actually to use to rebalance. This may be implicit in your thinking but the way you present it leads to possible confusion I think.

Consider the prototypical '60/40' investor. Even when the stock market drops 50%, and say based on long term history of zero correlation between treasuries and stocks, treasuries don't change (mainly for simplicity's sake, it doesn't fundamentally change the point), she only needs to liquidate 12 of the original 40 treasuries to rebalance to 60/40. And 12 is unlikely, the proportion is likely to be much lower in a given year say, whereas CD's might be maturing from a ladder every year. The rest of the fixed income will be left alone even with 'religious rebalance', and the mark to market gain recorded on the treasuries left alone, if so, would be a psychological advantage only. So I think everyone agrees that rebalancing using CD's is ill advised, but I'm not sure everyone is on the same page as to how little of the FI allocation this is likely to involve, and this has major implications.

For the generally large portion of FI which won't be used to rebalance, cash treasuries mean paying a lot for liquidity one doesn't need. Also the yield curve is the yield curve. If you ladder 5yr CD's to an eventual 2.5 yr average maturity of ladder in a stationary yield curve, you eventually receive 2.5% in 2.5 yrs, when the 2.5 yr treasury point is 165bps below that, much wider than the 5yr CD v 5yr treasury spread (not up to date numbers, but typical of recent times). That's 'roll yield' too, just recognized in a different way.

Also if 5yr CD's were 2.5% and treasury 1.5% (again broadly representative of recent times, not right now), the real spread is more like 1.25 ish considering the CD's put option, ie the present value of conditional cases when rates have gone up and you withdraw just to optimize that rate move, not to rebalance to stocks using CD's, which again everyone agrees CD's aren't suitable for.

And finally, whether or not one would personally use futures in their fixed income strategy, it's useful to refer to this strategy to show that the 'FDIC arb' is not just a liquidity phenomenon. For same 'generally true in recent times' numbers, treasury futures will give a return of the underlying ('cheapest to deliver') treasury issue for a particular contract minus the repo rate. But the repo rate is .25%-ish at most, while the cash one could put in an FDIC bank account for say 95% of the notional not required to be put up in cash for the futures would earn 1.1% at least. So that strategy would yield order of 75 bps more than a cash treasury, with no sacrifice in liquidity: withdrawal at will from the bank account, and treasury futures are more liquid and have lower transactions costs, either to 'roll' or cash in for rebalancing, than the 'off the run' treasury issues in which an individual would invest. There would only be direct credit exposure to the broker on ~1% actually required in cash margin, the rest of the 5% could be in other short term investments covered by SIPC held in the brokerage account ready to liquidate to generate additional margin as needed, and 95% is basically govt risk. And the 'roll yield' characteristics of the futures would be exactly the same as rolling the underlying treasuries. In fact the term 'roll yield' is from the futures world, 'harvesting the term premium (if any)' is actually what we mean more generally speaking.

In summary your point has merit, but only for the portion of FI likely or possible to be used in rebalancing in a given period, and that's a major qualification even if only CD's are considered as alternative, and is not a valid distinction at all between cash treasuries and FDIC cash/treasury futures method. For the large % of FI allocation which wouldn't or is very unlikely to be liquidated for rebalancing in a given period the advantage of best CD's over cash treasuries in recent yield environment is clear, and there's no real 'but' to it. But this advantage can vary, and in fact Friday 5yr treasury yield had risen to 1.74 and the best published 5yr CD rate on 'bankrate.com' has yet to rise from 2.27 (same best 5yr CD quote as a couple of weeks ago when the 5yr treas yild was only around 1.3).
Last edited by Johno on Sat Nov 07, 2015 4:16 pm, edited 2 times in total.

User avatar
Kevin M
Posts: 10002
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: CD 5-Year Report Card

Post by Kevin M » Sat Nov 07, 2015 4:09 pm

Doc wrote:
Kevin M wrote:The main point I think we are both making is that the perception that Treasuries will do well in a flight-to-quality scenario is exactly the reason, or at least a big reason, that Treasury rates are low and prices are high. This is the term risk piece coupled with extremely low default risk; i.e., long-term Treasuries (more term risk) did much better in 2008. The low default risk is key though, since long-term corporates did not do well in 2008. I don't think there is any possible disagreement here.
The disagreement is how you quantify the effect. You are quantifying it by looking at the "long" term difference in term risk and credit risk. I think there is also a very short term risk that is present only occasionally and it is a risk of not gaining more money rather than losing money.
I don't think we need to redefine risk here. Risk has an upside as well as a downside, otherwise there would be no point in taking risk. Clearly the potential upside inherent in a flight-to-quality episode is factored into Treasury pricing, which again, is a factor in why prices are high and interest rates are low relative to less safe bonds. I don't think you need to twist your brain into a pretzel to make risk seem like something that it's not.
During the Lehman crisis the price of one of the Vanguard Intermediate Term Treasury funds gained some 10% over a very short period of time. If such an occurrence happens once every 10 years and you capture half of it through re-balancing you have on average an additional 50 bps in favor of the Treasury.
First, assuming a financial crisis like 2008 every 10 years is ridiculous. It's widely agreed that 2008 was the worst financial crisis in the US and globally since the great depression. So based on history, perhaps once every 50-80 years would be a better assumption.

Next, why don't you extend your argument to long-term Treasuries, since these did by far the best in late 2008? Have you done a similar analysis for int-term vs. long-term Treasuries? Total returns for calendar 2008 (admiral shares):

Code: Select all

Int-Term:  13.49%
Long-Term: 22.69%
At the same time because of the EWP on the CD you have a disadvantage of maybe 10 to 20 bps. (You don't even get the 50% capture rate - It's 100%.)

We're starting to go in circles. We've already agreed that the potential rebalancing opportunity only lasted a few months, and that it may only require a relatively small portion of your fixed income to rebalance back to target. So, as I keep saying, hold a portion of your fixed income in long-term Treasuries if you want this potential benefit, and hold the rest in fixed income with less term risk, ideally with a yield premium available only to retail investors.
I assume since you are discussing a 5 year CD you are comparing it to a 10 year Treasury ladder.

Not really.

In part 1 of the blog post I compared a 5-year to a 5-year Treasury, since that's the most relevant comparison. In part 3, I compared these two to bond funds in 7 of the 9 cells in the 9-cell quality/maturity risk grid, with the objective of showing how return can be proportional to both types of risk (term/interest-rate and credit/default/quality) if the risks are rewarded (which it happened to be over the 5-year period examined). The main point was to help people understand why their bond fund might have done better or worse than another alternative, as well as highlighting the superior risk/return profile of a CD--especially a direct CD.

Those of us who have been participating in yield curve discussions for the last couple of years know very well that under the assumption of a static, positively sloped yield curve, a 10-year ladder will have almost exactly the same return as an individual 10-year bond held to maturity. So no, I would not say that comparing a 10-year Treasury ladder to a 5-year CD is a fair comparison.

Perhaps somewhat surprisingly, or perhaps not if you carefully examine yield curve changes, fund holding composition, etc., the Vanguard Intermediate-Term Treasury fund (admiral shares), which is roughly a 3-10 year Treasury ladder (with active yield curve management thrown in) had almost exactly the same 5-year return as the 5-year CD held to maturity (2.74% for CD, 2.73% for fund).
With the ladder you have the potential of roll yield that is not available from the CD again because of the EWP.
Although it's true that a marketable fixed-income security (CD as well as Treasury) has the potential of generating a price-gain component (I think the term "roll yield" has been horribly abused on this forum--just do a Google search on it) that can be harvested by selling the security, you get exactly the same effect if you hold a rolling CD ladder as you do from holding a rolling bond ladder, except you earn the extra yield premium offered by the CD.

You can even harvest the gain, if it occurs, from a brokered CD ladder, and although you may take a 1%-2% discount on the sale, if you can earn a 1% premium per year, you're ahead with the CD ladder if you can hold it for just one or two years before selling.

And once again, you can get the benefit, if it is realized, from holding part of your fixed income in a ladder of Treasuries, spanning whatever part of the yield curve you want to bet on for providing the price-gain benefit. Or just hold it in either the intermediate-term or long-term fund, depending on what you want to bet on.

Finally, the proof is in the pudding. Whatever price gain component was realized is included in the total return for the intermediate-term Treasury fund, which was basically the same as the 5-year CD held to maturity, which did not benefit from this effect.
Now if you don't re-balance religiously during a market crash a lot of this Treasury stuff goes away and Treasuries lose to the CD. But in this case Treasuries are the wrong comparison. If you don't care about now a total bond market fund or even an investment grade fund might suit your need better.
Once again, hold a combination of CDs and funds, like I do, but use a Treasury fund (unlike I do) if you want to maximize what you believe will be a benefit in a flight-to-quality event.

And use a long-term Treasury fund if you want to get more bang for the buck in this scenario. It occurs to me that this is similar to holding less equities but heavily tilted to small-value (the Larry Portfolio), or using small-cap value funds with larger loadings on the small and value factors to reduce the amount you have to put in higher cost funds.
The failure of many of these risk threads is that most retail investors even Bogleheads think of risk only as losing money. But the professionals who are in reality setting the market returns are also thinking of the risk of not gaining money.
Again, I think this is a poor way to frame it. The way academics frame it is that risk is the uncertainty of an investment earning its expected return over a specified holding period. Note that this doesn't say risk of losing money, since it incorporates upside uncertainty as well as downside uncertainty.

You may be right that most investors focus only on the downside, but after studying the risk/return topic extensively, I don't think about it this way (at least not when I'm thinking clearly). I admit that I tend to use examples that focus on the downside, since this seems to be what most people are concerned about, but I also try to emphasize that there's an upside to risk, and we happened to see that upside in the 5-year period examined in the blog post series.
Anyway good discussion.
Agreed!
Now I need to work on completing my 2015 tax return so that I can plan all my yearend tax gain harvesting.
Cool. Have done most of the tax-gain harvesting necessary for my fiancee to take advantage of her last year in the 15% federal tax bracket, but still have to sell a bit more to be sure we're close to the max. Also have to do a Roth conversion to get my federal tax up to $7,500, so I can get the full benefit of the electric vehicle tax credit for the Tesla I'm waiting to take delivery of in mid-December (which solved the problem of what to do with some of my CDs maturing this month). :beer

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

User avatar
Kevin M
Posts: 10002
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: CD 5-Year Report Card

Post by Kevin M » Sat Nov 07, 2015 5:42 pm

stlutz wrote:Just for kicks (it's cold out today), I made a up a bit of a backtest to try and simulate this over a longer period of time. I took the Simba spreadsheet and created a new data series I called "Bank CDs". I calculate each year's return by taking the average of the 5 year Treasury rates for the prior 5 years and then added 1.2% to the result to replicate a 5 year CD ladder. (So, right now I would be saying that new CDs are yielding 3%).
Nice addition to the conversation!
Results for the 1972-2014 period:

40% US stocks/20% Intl stocks/40% Int. Treasury: CAGR 9.79% St. Dev. 10.96
40% US stocks/20% Intl stocks/40% Bank CDs: CAGR 9.75% St. Dev. 11.00

In short, a wash. And you'll see why I picked 1.2% as the CD yield "bonus" over Treasuries--that was what I had to add to make the portfolio as good as one with Treasuries.
Interesting. I'd look at the following factors in evaluating the relevance of this backest, and in considering other backtests to evaluate:
  1. Historical time period covered.
  2. Stock/fixed asset allocation.
  3. Rebalancing policy.
  4. Using only one or the other fixed-income asset class vs. a combination.
With respect to factors 1, 2 and 3 I did a similar backtest almost exactly one year ago, but for 1953-2013 (longest period for which data was available for 5-year CMT in FRED), 100% fixed income, and the only "rebalancing" was to do early withdrawals from the CD only when a specified threshold was met:
The CD strategy is to do an early withdrawal and reinvest if the rate at the beginning of the year is greater than 0.5 percentage points over rate of the existing CD (so for a 2% rate, new rate would have to be greater than 2.5% to do an early withdrawal and reinvest). Otherwise, the CD is held for 5 years, and then a new CD purchased at the new rate. From 1954 to 2013, this resulted in doing early withdrawals only twice; in 1957 and 1970.
Also, I assumed only a 25 basis point yield premium for the CD, as opposed to the 100 basis point premium that has been about my average for CDs purchased over the last five years (actually somewhat higher, but ballpark). I posted the results here: 5-year CD vs. 5-year Treasury over 60 years - Bogleheads.

The post speaks for itself, so here I'll just note that not only did the CD strategy generate significantly higher returns, but also much higher risk-adjusted returns (much less return volatility of 5-year and 10-year returns).

Bringing factor 4 into the mix, it would be informative to do a similar backtest using a combination of CDs and Treasuries, perhaps even 10-year instead of 5-year for potentially higher flight-to-quality benefit, rebalancing only with Treasuries, but still taking advantage of the early withdrawal option for CDs, and to do this for a variety of stock/fixed ratios.

My portfolio is 30/70 stocks/fixed, so the fixed-income return is more significant, and the fixed income currently is 30/70 bonds/CDs, so there is plenty of bond space to use for rebalancing.

Another relevant observation is that Larry Swedroe, who is one of the most vocal proponents of looking at performance of the entire portfolio as opposed to individual components, uses CDs in his own portfolio as well as the portfolios of his firm's clients. Of course these probably are brokered CDs, so are used mainly for the yield premium over Treasuries, not for the added benefit of the early withdrawal option with direct CDs. I suspect that they are quite rational in deciding which fixed-income components to use in rebalancing decisions; of course they probably get better pricing on CDs in the secondary market than you or I would.

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

castlemodesto
Posts: 144
Joined: Sun Sep 15, 2013 12:31 pm

Re: CD 5-Year Report Card

Post by castlemodesto » Sat Nov 07, 2015 10:06 pm

Larry Swedroe ,a short while ago, in another thread ,posted his comment about CDs compared to treasuries and to TIPS
"first, TIPS relative to Treasuries look very cheap to me.
Second, why buy Treasuries though when can buy much higher yielding CDs without taking credit risk, at least out to 10 years. And compared to CDs TIPS no longer look cheap."

User avatar
Doc
Posts: 8510
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: CD 5-Year Report Card

Post by Doc » Sun Nov 08, 2015 10:02 am

We seem to be getting nowhere because we are lost in the minutia and we don't have the data anyway.

MY thesis is that their are other factors involved in the CD/Treasury comparison than just credit and term factors.

Some of these are:

a) The FDIC insurance on a CD is not the same as a Treasury. It doesn't cover future interest.

b) In a flight to quality situation Treasuries may perform better than CD's since CD's are not marked to market and have a withdrawal. I don't belive this is covered by what we usually think of as term risk.

c) The tax costs are different. If we have a 2% T-note and a 3% CD there is a 15 bps "penalty" that should be attributed to to the CD if your state tax rate is 5%.

d) There is a potential of roll yield with the Treasury because it is marked to market. Part of this may be covered in the term risk part. But there is also the potential benefit of an unexpected change in the yield curve. Even without an unexpected change there is the possibility of tax arbitrage by selling before maturity and thus changing your income from ordinary to LTCG.

In my opinion the 100 bps difference we see in the yield spreads is not a free lunch. Rather some if not most of it is due to other risk and tax factors that are not reflected in the two factor term + credit risk model. As I said up-thread I don't know how to quantify the factors. I can only guess and Kevin doesn't like some of my guesses like a 1 in 10 chance of Lehman type crisis in a given year. :(

At the risk of prolonging this thread further I came up with another thought early this morning. If we consider that part of our portfolio that is above any amounts needed for emergencies or crisis "abatement" would it not be better to compare the CD to a AA bond fund of similar duration instead of a Treasury. Information from such a comparison may also shed some light on my objections in the CD/Treasury methodology currently under discussion.
A scientist looks for THE answer to a problem, an engineer looks for AN answer and lawyers ONLY have opinions. Investing is not a science.

User avatar
Doc
Posts: 8510
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: CD 5-Year Report Card

Post by Doc » Sun Nov 08, 2015 10:16 am

Kevin M wrote:My portfolio is 30/70 stocks/fixed, so the fixed-income return is more significant, and the fixed income currently is 30/70 bonds/CDs, so there is plenty of bond space to use for rebalancing
We have a similar overall portfolio but I take a somewhat different approach than you suggest towards allocating the non re-balancing portion. I don't have either the knowledge or the time to try to maximize the risk adjusted return of the portfolio. I just take the Barcap Intermediate (1-10) Term Gov/Credit and try to mimic it with the re-balancing portion included. This slice and dice approach allows for more tax efficient placement of the slices and it allow me to use active funds in a lot of places that I normally wouldn't do.
A scientist looks for THE answer to a problem, an engineer looks for AN answer and lawyers ONLY have opinions. Investing is not a science.

Johno
Posts: 1883
Joined: Sat May 24, 2014 4:14 pm

Re: CD 5-Year Report Card

Post by Johno » Mon Nov 09, 2015 12:05 pm

Doc wrote:We seem to be getting nowhere because we are lost in the minutia and we don't have the data anyway.

MY thesis is that their are other factors involved in the CD/Treasury comparison than just credit and term factors.

Some of these are:

a) The FDIC insurance on a CD is not the same as a Treasury. It doesn't cover future interest.

b) In a flight to quality situation Treasuries may perform better than CD's since CD's are not marked to market and have a withdrawal. I don't belive this is covered by what we usually think of as term risk.

c) The tax costs are different. If we have a 2% T-note and a 3% CD there is a 15 bps "penalty" that should be attributed to to the CD if your state tax rate is 5%.

d) There is a potential of roll yield with the Treasury because it is marked to market. Part of this may be covered in the term risk part. But there is also the potential benefit of an unexpected change in the yield curve. Even without an unexpected change there is the possibility of tax arbitrage by selling before maturity and thus changing your income from ordinary to LTCG.

In my opinion the 100 bps difference we see in the yield spreads is not a free lunch. Rather some if not most of it is due to other risk and tax factors that are not reflected in the two factor term + credit risk model. As I said up-thread I don't know how to quantify the factors. I can only guess and Kevin doesn't like some of my guesses like a 1 in 10 chance of Lehman type crisis in a given year. :(
a) The risk is that the bank (but not the US govt) defaults and get back your principal but must find someplace else to put the money, which might have a lower rate than that CD. But it could have a higher rate. There's only a bias to a loss to the extent you accept the CD has superior value in the first place. And this all filtered through the probability of bank default, which is at most a few % as a reasonable ballpark. So this risk exists, but there's no plausible way it could be valued at more than a handful of bps.

b) this continues to ignore a key distinction at least two posters have made. The flight to quality increase in treasury prices, if so, is only objectively relevant to portions of the FI allocation one would sell in such circumstances. This is a small portion on a probability weighted basis. For the rest it's no actual advantage besides subjective money illusion type advantage. Or maybe the latter point needs to be isolated and debated. So this isn't a question of quantifying an across the board advantage, but of recognizing the basic point that the 'pro-CD' side is already accepting that CD's don't work for rebalancing; but you seem to resist engaging the point that most FI would not ever be used for rebalancing, especially not on a frequency less than annual maturities of a CD ladder.

c) this is understood, but state tax issues can't ever be used to draw general conclusions, because they vary so much. And 5% of the CD rate in a general recent comparison is 12.5bps out of 100 for 2.5 v 1.5.

d) As above, this is not correct. The curve is the curve. If it remains stationary upsloping, then there will be a term premium harvested by 'rolling' treasuries but will also be realized eventually in a different form by having a 2.5 yr maturity CD ladder which pays out the 5 yr rate. Moreover, 'roll yield' can be realized exactly the same way as with treasuries by rolling treasury futures while enjoying the ~75bp yield increase of funding the futures implicitly at the repo rate while investing the cash as best FDIC bank account rate. So it's also not a question of valuing 'roll yield' of treasuries v non roll yield of FDIC arb strategies (CD's or futures/bank acct): it's simply not correct to imagine that as unique to treasuries.

So it's not a matter of opinion actually. There is a state tax advantage to treasuries which depends on the state (zero in some) so it can't be generalized. Otherwise there's a plausible max few bps cost to the particular rate on the CD (though not its principle) being subject to the credit of the bank. Likewise there's a plausible few bps max cost to having ~1% of the principal at risk to the futures broker in futures/FDIC savings acct method (which also has a potentially significant tax advantage if the future can be 'stuffed' into tax deferred though the bank acct or treasuries would have to go in taxable). Then again 2.50 v 1.5 typical 5 yr comparison doesn't include the ~.25% value of the CD put, adds more to the 100 than the real value of anything you pointed out as chipping away at the 100. :D

'Free lunch' is a loaded term. The yield advantage of CD's makes them superior to 5 yr notes for any portion of the fixed income allocation which won't be used for rebalancing, and on a probability weighted basis that's most of most people's FI, for any plausible valuation of the other risk differences. The futures/cash method has also superior yield to treasuries without sacrificing any liquidity, relative to the marginal secondary risks, but does not give a cheap put option like direct CD, and requires labor input, which isn't strictly speaking a 'risk', but is a cost which varies relatively by individual. However it's clear that the FDIC spread is in fact 'too high' recently, not a matter of reasonable difference in opinion. The issue is mainly how to apply this premium in an advantageous way relative to liquidity needs.

User avatar
Kevin M
Posts: 10002
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: CD 5-Year Report Card

Post by Kevin M » Mon Nov 09, 2015 2:25 pm

Doc wrote: MY thesis is that their are other factors involved in the CD/Treasury comparison than just credit and term factors.
Johno did a better job than I could of debating the individual points.

I'd just like to point out that using credit risk and term risk as the dominant factors in evaluating overall fixed-income risk is not something I just pulled out of a hat. It is industry-standard and academic-standard, as far as I can tell. For example, both Vanguard and Morngingstar use a 9-cell quality/maturity grid, based on the two primary risk dimensions, to summarize the overall risk of fixed-income investments.

I'm not sure, but I believe the impetus of this was the Fama-French 5-factor paper, that added the term risk and credit risk factors to the small, value and market factors for stocks. Both Vanguard and Morngingstar, for example, use 9-cell grids to summarize risk for both stocks (Market-cap/style) and bonds (credit quality/maturity) that seem to stem from the FF research.

So although there certainly are other risk factors (Vanguard lists more than just two in its bond prospectuses), I think that it's widely accepted that credit risk and term risk are dominant, and explain the vast majority of the variability of returns when comparing nominal fixed income investments.

Fixed-income investors certainly should be aware of other factors, such as liquidity, but probably can best understand the relative risk of their nominal fixed-income investments by primarily using the two-dimensional model of credit risk and term risk.

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

User avatar
Doc
Posts: 8510
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: CD 5-Year Report Card

Post by Doc » Mon Nov 09, 2015 4:21 pm

I feel like my name is Ben. :D
Johno wrote:So this risk exists, but there's no plausible way it could be valued at more than a handful of bps.
In the present day probably yes. But during the S&L crisis IIRC there was considerable lost time involved before you got you principle returned. But agreed probably a small factor today.
Johno wrote:if so, is only objectively relevant to portions of the FI allocation one would sell in such circumstances. This is a small portion on a probability weighted basis.

I think I said the same thing a number of times. eg.
Doc wrote:If we consider that part of our portfolio that is above any amounts needed for emergencies or crisis "abatement" would it not be better to compare the CD to a AA bond fund of similar duration instead of a Treasury. Information from such a comparison may also shed some light on my objections in the CD/Treasury methodology currently under discussion.
Johno wrote:this is understood, but state tax issues can't ever be used to draw general conclusions, because they vary so much.
So the tax benefit of muni's is a figment on my youth?
Johno wrote:And 5% of the CD rate in a general recent comparison is 12.5bps out of 100 for 2.5 v 1.5.
Doc wrote:If we have a 2% T-note and a 3% CD there is a 15 bps "penalty" ...
So we don't take account of taxes but we do quibble on whether those non-taxes get applied to a 2.5% CD or a 3.0% CD?
Johno wrote:As above, this is not correct. The curve is the curve.
Doc wrote:But there is also the potential benefit of an unexpected change in the yield curve.
If the unexpected shift is in your favor you take advantage of it with Treasuries. With CD's you can't because the EWP wipes out most if not all of the gain. You also have the tax arbitrage possibility. But I forgot I'm not supposes to take taxes into account because people pay at different rates.

There are several differences in performances of CD's to Treasuries. Different investors are exposed to some of these and some are not. The importance of these differences depend on each investors tax rate, overall AA and overbalancing philosophy. To attribute the the entire 100 bps difference that we are seeing only to the difference in the FDIC limit seems naive to me. Unfortunately as this discussion has shown estimating the amount of the difference that is attributable to these other factors is difficult if not impossible to quantify.
A scientist looks for THE answer to a problem, an engineer looks for AN answer and lawyers ONLY have opinions. Investing is not a science.

User avatar
Doc
Posts: 8510
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: CD 5-Year Report Card

Post by Doc » Mon Nov 09, 2015 4:30 pm

Kevin M wrote:I'd just like to point out that using credit risk and term risk as the dominant factors in evaluating overall fixed-income risk is not something I just pulled out of a hat. It is industry-standard and academic-standard, as far as I can tell.
I'm an engineer who did all his undergraduate work with a slide rule not a computer spreadsheet. One thing that becomes apparent very rapidly from using such a tool is that when you subtract two similar size numbers each with a small error the error in the result is apt to be considerably larger. That's part of the issue here. A few additional factors even if small by themselves can have a large effect on the metric we are depending on.
A scientist looks for THE answer to a problem, an engineer looks for AN answer and lawyers ONLY have opinions. Investing is not a science.

Johno
Posts: 1883
Joined: Sat May 24, 2014 4:14 pm

Re: CD 5-Year Report Card

Post by Johno » Mon Nov 09, 2015 5:27 pm

Doc wrote:
Johno wrote:if so, is only objectively relevant to portions of the FI allocation one would sell in such circumstances. This is a small portion on a probability weighted basis.

1. I think I said the same thing a number of times. eg.
Doc wrote:If we consider that part of our portfolio that is above any amounts needed for emergencies or crisis "abatement" would it not be better to compare the CD to a AA bond fund of similar duration instead of a Treasury. Information from such a comparison may also shed some light on my objections in the CD/Treasury methodology currently under discussion.
2.
Johno wrote:this is understood, but state tax issues can't ever be used to draw general conclusions, because they vary so much.
So the tax benefit of muni's is a figment on my youth?
3.
Johno wrote:As above, this is not correct. The curve is the curve.
Doc wrote:But there is also the potential benefit of an unexpected change in the yield curve.
If the unexpected shift is in your favor you take advantage of it with Treasuries. With CD's you can't because the EWP wipes out most if not all of the gain.
4. To attribute the the entire 100 bps difference that we are seeing only to the difference in the FDIC limit seems naive to me. Unfortunately as this discussion has shown estimating the amount of the difference that is attributable to these other factors is difficult if not impossible to quantify.
1. But then I don't understand why you also give liquidity or 'flight to quality price bump' seemingly as if a factor systematically in favor of treasuries v CD's. Rather, to simplify the probability distribution, it means CD's just aren't suitable for the portion of FI 'likely' to be used to rebalance, but OTOH liquidity makes no dent in the CD's yield advantage for the portion for which its 'not likely'. Of course actually it's probability continuum considering the volatility for stocks and the interval of time between rebalancings, but for most of the FI portfolio of most investors, there's no benefit to liquidity in terms of rebalancing, and the only really valid, systematic* argument for treasuries v CD's goes out the window on that latter portion.

It goes out the window, period, if one would consider the treasury futures plus bank account technique.

AA bonds OTOH have credit risk on the principal. There's no more reason to consider them in lieu of CD's than in lieu of treasuries. That's mainly a non sequitur. And the need for credit diversification and high bid-offer of non-treasury bonds generally forces retail investors to buy *funds* of IG bonds where yield leaks away in a variety of ways which don't apply to either CD's or treasuries.

*2. Again I take it as given everyone knows CD's are subject to state tax (if any) and treasuries not. There's really no point in dwelling on it, and eg. in Florida or Texas that means no difference, nor in any state if in a tax deferred account. And it's not a significant % of the yield difference in most cases even in a taxable account.

3. This misunderstanding, common on this generally knowledgeable forum but no less a misunderstanding for it, seems to stem from failing to recognize the mismatch to begin with in comparing a constant maturity investment (a treasury fund say) from a single instrument you hold to maturity. A single 5 yr note you hold to maturity also has exactly the same 'roll yield' characteristic as a CD you hold to maturity. Likewise a ladder of treasuries with a given avg maturity has the same 'roll yield characteristic' as a similar ladder of CD's. You can't practically mimic with CD's a treasury position where you buy instruments at Avg mat+.25yr and sell them at Avg mat-.25yr, but that's not the only portfolio on which a term premium can be harvested. It's not correct to state the foregoing facts as 'CD's don't have a roll yield'. And again anyway if the investor is convinced of the prospects of realizing an attractive term premium with tightly clustered maturities, buy and sell...the most efficient way to do that is the place where the term 'roll yield' comes from in the first place, buy eg. the Dec 5 yr note futures now, sell it for the Mar 2016 around the end of this month, and so on qtrly, and put the cash in a bank account yielding 1.1% v ~.25% implied financing rate on the treas futures.

4. Again the futures/bank acct pricing relationship shows that a lot of the FDIC spread is not actually liquidity, because the liquidity there >=cash treasuries but yield pick up still order of 75bps (and as for taxes that method is potentially superior to either of the other two).

It's knocking over a straw man to argue against every last bp being 'for free', nobody is saying that, just pointing out the implausibility of 'guessing' that 75-100 bps all gets eaten up by taxes (in any normal situation) or by other risks: that's pretty clearly not the case. For the portion of fixed income not needed or unlikely to be needed for rebalancing, most of most investor's FI, the CD is simply superior to cash treasury for retail investors in recent pricing relationship. For those willing to use it, futures/cash combination is also superior in yield to cash treasuries out of proportion to minor risk difference, and fully liquid.

User avatar
Kevin M
Posts: 10002
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: CD 5-Year Report Card

Post by Kevin M » Mon Nov 09, 2015 6:15 pm

Again, Johno is knocking it out of the park, so I'll just kick back and let him continue to explain the finer points.

I would like to reply to this one comment though, especially since it's been quoted a couple of times in replies:
Doc wrote:At the risk of prolonging this thread further I came up with another thought early this morning. If we consider that part of our portfolio that is above any amounts needed for emergencies or crisis "abatement" would it not be better to compare the CD to a AA bond fund of similar duration instead of a Treasury. Information from such a comparison may also shed some light on my objections in the CD/Treasury methodology currently under discussion.
I'm not even sure I understand this comment, because it seems like such a stretch to suggest that a federally-insured CD, especially one with a good early withdrawal option, is comparable to an AA-rated bond fund in terms of term risk or credit risk.

This actually highlights a point I want to make: the blog post series isn't about a "CD/Treasury methodology"--it's about using the standard credit-risk/term-risk framework to help understand the variation in returns of various fixed-income instruments, including CD, Treasury, Treasury funds with different average maturities, AA bond funds with different average maturities, and a B bond fund with medium term average maturity.

I just happen to think that an individual Treasury of same maturity is the best benchmark against which to compare a CD:
  • For any kind of CD, the credit risk is essentially the same for the retail investor who can stay within federal-insurance limits.
  • For a brokered CD, term risk also is comparable, except the Treasury has an edge because of liquidity. However, if the CD yield premium is rich compared to the relative illiquidity, as it often has been in recent years, it may be well worth it to the investor who can afford to hold for one or two years before selling.
  • A direct CD has less term risk because of the early withdrawal penalty (acknowledging that term risk has a potential upside as well as downside).
Once you start comparing to bond funds, things get more complicated, because you now are comparing a rolling ladder to a single security, so you really should compare to a comparable rolling ladder of CDs. Start with brokered CDs to make it easier to comprehend that "roll yield" is the same for both, keeping yield and liquidity differences in mind. Then extend to direct CDs, with the understanding that although the "roll yield" cannot be harvested in the same way, it still exists.

An AA bond fund clearly has more credit risk, and any bond fund has more term risk than a direct CD with a good early withdrawal option. I don't know how anyone could argue that they have similar risk.

We actually can see an overview of how the market is pricing credit risk and term risk by looking at the Fidelity table of yields/rates for bonds with different maturities and quality ratings, along with CDs: CUSIP Lookup and Bond Yields - Fidelity. Vanguard also provides something similar, but without as fine a gradation of quality ratings: Vanguard: Find bonds and CDs.

Although we see CD yields falling somewhere between corporate AA and corporate AAA yields out to maturity of 10 years, this makes sense if we think of it as institutional pricing, since CDs are not risk-free to the institutional investor.

Also note that at 20-year maturity, even Agency/GSE securities have higher yield than CDs, and corporate AA yield is much higher (almost 5%, compared to 3% for CD). So the "CD yield curve" is flat between 10 and 20 years, very unlike investment-grade corporate bonds. Apparently the 20-year CD is priced as much less risky than even a AAA corporate bond.

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

User avatar
Doc
Posts: 8510
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: CD 5-Year Report Card

Post by Doc » Tue Nov 10, 2015 9:52 am

Kevin M wrote:Doc wrote:
At the risk of prolonging this thread further I came up with another thought early this morning. If we consider that part of our portfolio that is above any amounts needed for emergencies or crisis "abatement" would it not be better to compare the CD to a AA bond fund of similar duration instead of a Treasury. Information from such a comparison may also shed some light on my objections in the CD/Treasury methodology currently under discussion.

I'm not even sure I understand this comment, because it seems like such a stretch to suggest that a federally-insured CD, especially one with a good early withdrawal option, is comparable to an AA-rated bond fund in terms of term risk or credit risk.
Fitch ratings:
Some examples of letter ratings include:
AAA - reliable and stable
AA - quality with a bit higher risk

A quick M* screen for AA rated index funds produced a double handful of funds with durations of a little over five years and average yield (eyeball) of ~2.3%. The two Vg TBM funds are in the list. The current yield on the five is ~1.7%. So based on difference between the AA funds and the five the AAA to AA premium should be about 60 bps. Yet we are seeing closer to 100 between the CD and the five. Something is going on here that I can't explain using only credit and term risk in the model.
A scientist looks for THE answer to a problem, an engineer looks for AN answer and lawyers ONLY have opinions. Investing is not a science.

Post Reply