Why GNMAs Shouldn't Be Your Bond Choice

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
User avatar
tc101
Posts: 3014
Joined: Tue Feb 20, 2007 3:18 pm
Location: Atlanta - Retired in 2004 at age 54

Why GNMAs Shouldn't Be Your Bond Choice

Post by tc101 » Sat Feb 03, 2018 10:46 am

Here is an article by Larry Sedroe back in 2010.

https://www.cbsnews.com/news/why-gnmas- ... nd-choice/

At that time he recommended against holding GNMAs, and even recommended against the total bond market fund because it held them. I read a lot about this on Bogleheads back about the time the article was written. I haven't seen the idea mentioned here in years. What are your thoughts?
. | The most important thing you should know about me is that I am not an expert.

UpperNwGuy
Posts: 1107
Joined: Sun Oct 08, 2017 7:16 pm

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by UpperNwGuy » Sat Feb 03, 2018 10:59 am

I've seen several recent posts mention mortgage based securities. Usually MBS is mentioned when comparing the Vanguard Total Bond Market Index Fund (which has MBS) to the Vanguard Intermediate Term Bond Index Fund (which does not have MBS). Some of those who post dislike MBS, while others appear to be indifferent as to their presence in Total Bond. I don't think I've seen anyone get truly excited about MBS.

A similar topic that seems to be ignored for long periods of time is that of asset-backed securities. ABS show up in a lot of Vanguard bond mutual funds. Seems to me that they would have some of the same issues as MBS.

Disclosure: I own Total Bond.

User avatar
Blueskies123
Posts: 437
Joined: Sat Nov 15, 2014 7:18 pm
Location: South Florida

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by Blueskies123 » Sat Feb 03, 2018 11:14 am

I also noticed there is very little discussion about GNMAs. I have had them for probably 20 years in my bond portfolio. I have read the Wiki about posting pictures but I still find it cumbersome to snap a pic from Portfolio Visualizer. PV only had 10 years of data but the GNMA had the best Sortino Ratio

https://www.portfoliovisualizer.com/bac ... ion3_3=100
Last edited by Blueskies123 on Sat Feb 03, 2018 11:40 am, edited 1 time in total.

montanagirl
Posts: 973
Joined: Thu Nov 19, 2009 4:55 pm
Location: Montana

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by montanagirl » Sat Feb 03, 2018 11:20 am

How did GNMA fare during the mortgage meltdown 2007-2009?

It didn't seem like the GNMA fund was affected that much.

livesoft
Posts: 62776
Joined: Thu Mar 01, 2007 8:00 pm

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by livesoft » Sat Feb 03, 2018 11:20 am

Blueskies123 wrote:
Sat Feb 03, 2018 11:14 am
I also noticed there is very little discussion about GNMAs. ...
On bogleheads.org various fads come and go. There has been in the past significant discussion about GNMAs, but it is true there hasn't been much, if any, discussion lately.

I consider GNMAs government-backed and not MBS. I consider MBS something backed by a place like the ol' Countrywide folks.
Wiki This signature message sponsored by sscritic: Learn to fish.

tibbitts
Posts: 8044
Joined: Tue Feb 27, 2007 6:50 pm

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by tibbitts » Sat Feb 03, 2018 11:26 am

Although we shouldn't put too much emphasis on performance, I have to admit I've been disappointed by recent GNMA performance for a little while now, but when I say disappointed I mean by fractions of a percent per year. I think that for all the noise about the theoretical problems with GNMA, the worst case is you lose a small fraction of a percent per year vs. reasonable alternatives, and the best case is you gain a few - it's just that the former seems to have happened somewhat surprisingly (to me) more often than the latter.

User avatar
triceratop
Moderator
Posts: 5717
Joined: Tue Aug 04, 2015 8:20 pm
Location: la la land

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by triceratop » Sat Feb 03, 2018 11:26 am

montanagirl wrote:
Sat Feb 03, 2018 11:20 am
How did GNMA fare during the mortgage meltdown 2007-2009?

It didn't seem like the GNMA fund was affected that much.
GNMAs do not have credit premia because they are backed by the full faith and credit of the US government. This is different than the MBS products you heard about in e.g. The Big Short
"To play the stock market is to play musical chairs under the chord progression of a bid-ask spread."

User avatar
munemaker
Posts: 3503
Joined: Sat Jan 18, 2014 6:14 pm

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by munemaker » Sat Feb 03, 2018 2:11 pm

Personally I think the Vanguard GNMA fund is a good choice for your bond allocation. Before I became aware of the Bogleheads forums, I used GNMA as my bond allocation for decades. If you compare on Morningstar, you will see the long term performance relative to Total Bond Market (VBTLX) is nearly identical to Vanguard GNMA (VFIJX). VBTLX contains a portion of corporate bonds that are not government backed, so it would have more credit risk, compared to GNMAs which are federally backed.

For example:
For 15 years returns: VBTLX = 3.99%, VFIJX = 3.96%
For 10 year returns: VBTLX = 3.57%, VFIJX = 3.62%
For 5 year returns: VBTLX = 1.83%, VFIJX = 1.71%

Pretty much even. I do remember someone on here saying the problem with GNMAs is when interest rates fall, people will refinance their mortgages and this creates a negative bias of sorts in the returns. All I can say is the rates seem to be competitive with Total Bond Market.

Why didn't I stick with GNMAs? Well, I drank the Total Bond Market cool aid on this site. It doesn't seem to really make much difference either way.
Last edited by munemaker on Sat Feb 03, 2018 2:13 pm, edited 1 time in total.

Valuethinker
Posts: 36314
Joined: Fri May 11, 2007 11:07 am

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by Valuethinker » Sat Feb 03, 2018 2:12 pm

livesoft wrote:
Sat Feb 03, 2018 11:20 am
Blueskies123 wrote:
Sat Feb 03, 2018 11:14 am
I also noticed there is very little discussion about GNMAs. ...
On bogleheads.org various fads come and go. There has been in the past significant discussion about GNMAs, but it is true there hasn't been much, if any, discussion lately.

I consider GNMAs government-backed and not MBS. I consider MBS something backed by a place like the ol' Countrywide folks.
You are kidding, right? You understand the difference?

A Mortgage Backed Security is an MBS if the underlying asset pools is made up of mortgages, regardless of who issues it. It is the securitization process that makes it an MBS.

It so happens the US Treasury guarantees the GNMAs. A Canadian MBS is normally guaranteed by the Canadian Mortgage and Housing Corporation (CMHC).

However you still have the issues of US mortgages being repayable ahead of maturity, or outside the lock in period (typically 5 years in other countries, when you can then refinance, but you cannot do so, except with big penalties, before the fixed rate period expires).

Thus you still have prepayment risk or extension risk, which gives rise to the property of negative convexity in the bonds.

Valuethinker
Posts: 36314
Joined: Fri May 11, 2007 11:07 am

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by Valuethinker » Sat Feb 03, 2018 2:17 pm

montanagirl wrote:
Sat Feb 03, 2018 11:20 am
How did GNMA fare during the mortgage meltdown 2007-2009?

It didn't seem like the GNMA fund was affected that much.
GNMA bonds, and those of FHA, Tennessee Valley Authority and some other US government agencies are backed by the full faith and credit of the US government.

FNMA and FMAC were privatized in the 1960s and 1970s, those bonds did not have that legal protection under law.

In the summer of 2008 the Bush Administration took those 2 entities into government "conservatorship". Thus, although there is no de jure guarantee of the bonds, the market treats them as a de facto guarantee of creditworthiness-- there's no significant difference in yield between those 2 companies and the legally guaranteed bonds. The 2 companies now pay substantial dividends to the US Treasury, and as a result have depleted their capital-- there has been, as yet, no agreement between President & Congress as to what happens when they can no longer issue bonds.

The bonds in The Big Short were often "private label" Mortgage Backed Securities, i.e. not issued by any of the Agencies (GNMA, FMAC & FNMA plus the others). These were often high risk "non conforming" mortgages that the agencies, legally, could not buy & securitize.

In addition, when they were still private cos, FNMA and FMAC got involved in the creation of more complex financial structures (CDOs, CDO squareds, etc.; "synthetic" CDOs) which eventually did show significant losses.

User avatar
nisiprius
Advisory Board
Posts: 36845
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by nisiprius » Sat Feb 03, 2018 5:45 pm

I don't have any stake in this, apart from owning Total Bond which includes some GNMAs. Indeed Larry Swedroe has recommended against Total Bond for that reason and says so in that article. I would never personally dream of going 100% GNMA, but only because it feels undiversified, but the only reason is that I don't have the expertise to form a judgement on whether it would be good, bad, or indifferent to concentrate into one single category of bond.

The Vanguard GNMA fund, VFIIX was launched in 1980 and thus antedates Total Bond by six years. I have the impression that some VFIIX holders bought it before Total Bond was launched, as serving much the same purpose, and kept it because they have seen no reason to change.

Despite Larry Swedroe's expertise and all of his explanations of the potential risks in GNMAs, they've always sounded like, at worst, "risk of slight underperformance," ego risk rather than serious financial risk. Whenever I stare at the growth charts I sure don't see any place in thirty-seven years where the alleged potential risk has ever showed up.

To the eyeball,
Source
Image

there's no obvious difference in pattern between the two other than VFIIX having slightly higher return. PortfolioVisualizer is showing VFIIX as having slightly higher return combined with slightly lower risk (as measured by standard deviation, worst year, and maximum drawdown).

I mean, I don't know. I completely fail to see any place where VFIIX has ever had a problem, and I'm not clear on how big a problem Larry Swedroe thinks there might be.

And I'm certainly not going to fret about a quarter of my Total Bond in a bond category that I can't actually see there ever having been any problems in.

But if you don't like GNMAs for any reason, it's easy enough to find Vanguard bond funds that don't include them.
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.

lack_ey
Posts: 6619
Joined: Wed Nov 19, 2014 11:55 pm

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by lack_ey » Sat Feb 03, 2018 5:58 pm

Just keep in mind that Vanguard's GNMA fund is actively managed and I'm pretty sure has outperformed historically (certainly compare to the GNMA ETF the last 6 years). It's maybe not the fairest way to assess relative performance unless you think the fund alpha is persistent.


In any case, as to the broader question, I don't have a strong opinion on MBS generally or GNMAs specifically. I have some trouble writing off such a large part of the bond market as significantly mispriced, though, or inappropriate for individual investor asset allocations, without some good justification. Sure, we know about prepayment risk, the negative convexity, the fact that prepayment risk didn't show up much during the most recent financial crisis because of issues specific to that one (people not being able to refinance) and everything else, all the potential downsides. That's all priced in, and accordingly, unless it's priced in significantly incorrectly, then I don't see it making a big difference.

stlutz
Posts: 4780
Joined: Fri Jan 02, 2009 1:08 am

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by stlutz » Sat Feb 03, 2018 6:57 pm

Despite Larry Swedroe's expertise and all of his explanations of the potential risks in GNMAs, they've always sounded like, at worst, "risk of slight underperformance," ego risk rather than serious financial risk. Whenever I stare at the growth charts I sure don't see any place in thirty-seven years where the alleged potential risk has ever showed up.
In fairness, I think it's in a 1970s scenario where the risk has "shown up". When rates are increasing that dramatically, you can get in a situation where a homeowner can't afford to move and thus the bonds experience that extension risk.

I was looking for a GNMA fund that had been around that long. The Franklin US Government Securities fund (FKUSX" was one example I could find. Looking at M* is does appear that it did noticeably worse (i.e. I can visually see the difference on a chart) than other government bonds in the late 70s. LEXNX--Voya GNMA Income Fund was another example. Same pattern.

jginseattle
Posts: 725
Joined: Fri Jul 01, 2011 7:33 pm

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by jginseattle » Sat Feb 03, 2018 8:13 pm

I avoid GNMAs. In addition to Mr. Swedroe, this is also what David Swensen recommends doing.

It's very difficult to model prepayments, for example.

User avatar
telemark
Posts: 2317
Joined: Sat Aug 11, 2012 6:35 am

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by telemark » Sat Feb 03, 2018 8:26 pm

The referenced article says
You get to collect the risk premium only if rates are relatively stable
and rates have been pretty stable since 2010. So the problem, if there is one, hasn't manifested itself recently.

User avatar
patrick013
Posts: 2403
Joined: Mon Jul 13, 2015 7:49 pm

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by patrick013 » Sat Feb 03, 2018 9:00 pm

GNMA's when rates are high can be a good investment bought
at a discount and held to maturity. Those prepayments can
happen anytime and usually shorten the maturity quite well.
Safe and good for bond diversification somewhat. I'd buy
new issues.

Agency Bonds
age in bonds, buy-and-hold, 10 year business cycle

User avatar
nedsaid
Posts: 10482
Joined: Fri Nov 23, 2012 12:33 pm

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by nedsaid » Sat Feb 03, 2018 9:46 pm

Valuethinker wrote:
Sat Feb 03, 2018 2:12 pm
livesoft wrote:
Sat Feb 03, 2018 11:20 am
Blueskies123 wrote:
Sat Feb 03, 2018 11:14 am
I also noticed there is very little discussion about GNMAs. ...
On bogleheads.org various fads come and go. There has been in the past significant discussion about GNMAs, but it is true there hasn't been much, if any, discussion lately.

I consider GNMAs government-backed and not MBS. I consider MBS something backed by a place like the ol' Countrywide folks.
You are kidding, right? You understand the difference?

A Mortgage Backed Security is an MBS if the underlying asset pools is made up of mortgages, regardless of who issues it. It is the securitization process that makes it an MBS.

It so happens the US Treasury guarantees the GNMAs. A Canadian MBS is normally guaranteed by the Canadian Mortgage and Housing Corporation (CMHC).

However you still have the issues of US mortgages being repayable ahead of maturity, or outside the lock in period (typically 5 years in other countries, when you can then refinance, but you cannot do so, except with big penalties, before the fixed rate period expires).

Thus you still have prepayment risk or extension risk, which gives rise to the property of negative convexity in the bonds.
Extension risk and prepayment risk are a risk for ALL bonds save for US Government Bonds and perhaps sovereign debt. A bond issuer will do what a mortgage holder will do when interest rates drop, that is refinance. This is why many bonds have call dates. On the other hand, if interest rates increase, bond issuers are not going to call their bonds and refinance at higher interest rates.
US Treasury Debt is not callable, I don't know about the sovereign debt of other nations. My suspicion is that sovereign debt is not callable either.

I would say that prepayment risk and extension risk is higher for Mortgage Backed Securities than it is for other bonds. How much higher is a matter of debate. When interest rates were cranked up by Paul Volker to crush inflation, the value of mortgages held in bank portfolios got crushed and thus much of the Savings and Loan Industry in the United States had negative equity. There are unique risks to ALL classes of bonds and there is a scenario out there that could crush even the "safest" of bonds. But these scenarios have been unlikely and most often short lived.

Two examples I can think of when I discuss unique risks are money market funds and TIPS. The credit markets got so panicked in the 2008-2009 financial crisis that AAA credit General Electric could not roll over its commercial paper. Likewise, the market for other short term credit also dried up and money market funds faced the prospect of breaking the buck or showing the Net Asset Value of these funds to be less than $1.00. Who would have thought there would be what was in effect a run on the money market funds. Because of the quality and short-term nature of the instruments bought by money market funds, these funds were thought to be very safe until the 2008-2009 crisis, Uncle Sam had to step in to slap guarantees on these funds.

TIPS are a much, much smaller market than nominal treasuries. Though these are US Treasury Instruments, they were not as liquid as nominal treasuries and they dropped 11-12% during the crisis.

Who would have thought that money market funds and TIPS could be so risky. Well, that unique situation that exposed the unique risks of these instruments happened and "safe" investments turned risky. GNMAs and other such government agency bonds performed very well during the crisis and the unique risks of these instruments didn't show up during 2008-2009. My suspicion is that Mortgage Backed Securities got crushed in the late 1970's and early 1980's. This time period was not good for other classes of bonds either. Over the long term, instruments such as GNMAs have been good investments but they do have their risks.
A fool and his money are good for business.

NibbanaBanana
Posts: 224
Joined: Sun Jan 22, 2017 10:34 pm

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by NibbanaBanana » Sat Feb 03, 2018 11:07 pm

If there are as many smart cookies and rocket scientists out there pricing bonds as I've heard, then why wouldn't they be able to price in the extra risk? Why does the 32 year chart above look like the GNMA fund compares favorably to total bond over a long period if there's more risk? Zoom in and observe the behaviour during the GFC. Looks like GNMA held up better than total bond. Why would VG include it in their STAR fund if they didn't consider it one of their best bond funds?

I've read about the risks particular to GNMA's in John bogle's book. But these questions still remain in my mind.

jalbert
Posts: 3806
Joined: Fri Apr 10, 2015 12:29 am

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by jalbert » Sun Feb 04, 2018 12:25 am

The total bond market index is about 20% residential MBS but only a tiny fraction is GNMAs. People assume FNMAs and FRMCs are fully backed by the govt, but that is incorrect. The prospectus for fannie and freddie passthroughs explicitly states that they are not govt backed.

Corporate bonds have prepayment risk and extension risk as well as credit risk. GNMAs should be less correlated with corporate bonds than other MBS would be.

Individual corporate bonds are much riskier than individual GNMAs but credit risk is diversifiable. Prepayment risk and extension risk of GNMAs may be dealt with by active management. For instance, a simple point is that VA loans are assumable and don't have to be paid off when a house with one is sold. Extension risk is much higher for VA loans than FHA loans as a result. A GNMA index fund has to track the index but an actively managed GNMA portfolio can minimize VA mortgage exposure.

If you run a risk parity portfolio optimization with intermediate treasuries, GNMAs, corporate bonds, and short-term TIPs a common result is something like:

Intermediate treasuries 23%
Corporate bonds 23%
GNMAs 36%
Short-term TIPs 18%

You may not want that high of an allocation, but GNMAs are effective bond portfolio diversifiers. One could implement such a portfolio with VBILX, VTAPX, VFIIX.
Risk is not a guarantor of return.

Valuethinker
Posts: 36314
Joined: Fri May 11, 2007 11:07 am

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by Valuethinker » Sun Feb 04, 2018 9:23 am

nedsaid wrote:
Sat Feb 03, 2018 9:46 pm
Valuethinker wrote:
Sat Feb 03, 2018 2:12 pm
livesoft wrote:
Sat Feb 03, 2018 11:20 am
Blueskies123 wrote:
Sat Feb 03, 2018 11:14 am
I also noticed there is very little discussion about GNMAs. ...
On bogleheads.org various fads come and go. There has been in the past significant discussion about GNMAs, but it is true there hasn't been much, if any, discussion lately.

I consider GNMAs government-backed and not MBS. I consider MBS something backed by a place like the ol' Countrywide folks.
You are kidding, right? You understand the difference?

A Mortgage Backed Security is an MBS if the underlying asset pools is made up of mortgages, regardless of who issues it. It is the securitization process that makes it an MBS.

It so happens the US Treasury guarantees the GNMAs. A Canadian MBS is normally guaranteed by the Canadian Mortgage and Housing Corporation (CMHC).

However you still have the issues of US mortgages being repayable ahead of maturity, or outside the lock in period (typically 5 years in other countries, when you can then refinance, but you cannot do so, except with big penalties, before the fixed rate period expires).

Thus you still have prepayment risk or extension risk, which gives rise to the property of negative convexity in the bonds.
Extension risk and prepayment risk are a risk for ALL bonds save for US Government Bonds and perhaps sovereign debt. A bond issuer will do what a mortgage holder will do when interest rates drop, that is refinance.
Just to amend that:

- generally the Mortgaged Backed Securities of other countries do not have prepayment (or extension) risk because the phenomenon of the 25+ year fixed rate mortgage (that is refinanceable) does not exist
- corporate bonds are often issued without call provisions. Any prepayment then would take place by the issuer buying back the bonds in the market, at a premium, and they usually cannot force the holder to sell
This is why many bonds have call dates. On the other hand, if interest rates increase, bond issuers are not going to call their bonds and refinance at higher interest rates.
Many bonds do not have call dates.
US Treasury Debt is not callable, I don't know about the sovereign debt of other nations. My suspicion is that sovereign debt is not callable either.
Up until 1986 US Treasury Bonds were callable. The UK has had some callable issues. AFAIK sovereign bonds are generally not callable.
I would say that prepayment risk and extension risk is higher for Mortgage Backed Securities than it is for other bonds. How much higher is a matter of debate. When interest rates were cranked up by Paul Volker to crush inflation, the value of mortgages held in bank portfolios got crushed and thus much of the Savings and Loan Industry in the United States had negative equity. There are unique risks to ALL classes of bonds and there is a scenario out there that could crush even the "safest" of bonds. But these scenarios have been unlikely and most often short lived.
When your coupon is low, your duration is higher. It takes a smaller move in interest rates to achieve the same price change in the bond.
Two examples I can think of when I discuss unique risks are money market funds and TIPS. The credit markets got so panicked in the 2008-2009 financial crisis that AAA credit General Electric could not roll over its commercial paper.
It was not just GE, it was all corporates. I don't think GE was still AAA at that point?
Likewise, the market for other short term credit also dried up and money market funds faced the prospect of breaking the buck or showing the Net Asset Value of these funds to be less than $1.00. Who would have thought there would be what was in effect a run on the money market funds. Because of the quality and short-term nature of the instruments bought by money market funds, these funds were thought to be very safe until the 2008-2009 crisis, Uncle Sam had to step in to slap guarantees on these funds.
The liquidity crisis though was global, the collapse of the SIVs/ Conduits had a similar effect.
TIPS are a much, much smaller market than nominal treasuries. Though these are US Treasury Instruments, they were not as liquid as nominal treasuries and they dropped 11-12% during the crisis.

Who would have thought that money market funds and TIPS could be so risky.
The TIPS were being used as collateral for various REPO transactions. When Lehman failed, it led to a forced sale of collateral by the collateral holders (since Lehman was not going to pay them back the money Lehman had been lent).

TIPS were always illiquid amongst government bonds. Most holders of TIPS hold to maturity. Thus the selling had a disproportionate impact on price.
Well, that unique situation that exposed the unique risks of these instruments happened and "safe" investments turned risky. GNMAs and other such government agency bonds performed very well during the crisis and the unique risks of these instruments didn't show up during 2008-2009.
FNMA and FMAC bonds had plunged earlier in the summer, until the US Treasury put the 2 firms into conservatorship. In the flight to safety, GNMA bonds were an example of a US government backed security, therefore safe, therefore buyable. In addition, official interest rates were falling very fast, even if LIBOR was not. That was good for all risk-free bonds.

My suspicion is that Mortgage Backed Securities got crushed in the late 1970's and early 1980's. This time period was not good for other classes of bonds either. Over the long term, instruments such as GNMAs have been good investments but they do have their risks.
The assumptions about repayment were much less in that time frame. It was expected that people would hold their mortgages, without refinancing, for much longer periods of time, thus the maturity of the securities was much later. At least that is my memory of what was the case.

So the spikes in interest rates did not have a disproportionate impact on the prices of GNMAs v. US Treasury Bonds-- bad for both, but not worse for GNMAs (much).

The mathematics of managing portfolios of of MBS, the tools, were not as sophisticated then. It was Lew Ranieri at Salomon who drove this market in the late 70s/ early 80s. Tools like you have on a Bloomberg terminal were just not available, then-- in fact Michael Bloomberg got his start after being fired as an equity trading partner of Salomon in, from memory, 1979.

Valuethinker
Posts: 36314
Joined: Fri May 11, 2007 11:07 am

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by Valuethinker » Sun Feb 04, 2018 11:45 am

NibbanaBanana wrote:
Sat Feb 03, 2018 11:07 pm
If there are as many smart cookies and rocket scientists out there pricing bonds as I've heard, then why wouldn't they be able to price in the extra risk? Why does the 32 year chart above look like the GNMA fund compares favorably to total bond over a long period if there's more risk? Zoom in and observe the behaviour during the GFC. Looks like GNMA held up better than total bond. Why would VG include it in their STAR fund if they didn't consider it one of their best bond funds?

I've read about the risks particular to GNMA's in John bogle's book. But these questions still remain in my mind.
There is a piece in a footnote (p 165 from memory but I don't have a copy to hand) to Richard Bookstaber's "A Demon of our own Design" about the trading risks that the big investment bank trading desks were running on MBS-- re the negative convexity problem. Bookstaber is one of the real risk experts out there, ran the risk desk for Morgan Stanley for a few years and has written pre and post crash about the subject.

In fact, that risk never showed up, it was a solvency collapse (on the part of borrowers) that created the Great Financial Crisis.

Markets are probabilistic. If you tell me the future volatility of a security with embedded optionality (like an MBS) I can tell you the right price.

This is measures by the Option Adjusted Spread-- the additional yield available over the comparable US Treasury bond reflecting early repayment or extension risk, that is measured by the negative convexity of the bond* (assuming both bonds have the same credit risk, as they would for a GNMA; there is also more liquidity risk in the MBS).

However that probability can change. If something has a payoff of either 1 or 0 with 50% probability each, then the Expected Value = 50% x 1 + 50% x 0 = 0.5 (that's the basis of option pricing, the binomial model).

But that's ex ante. Ex post, you either made 1 or 0 out of holding the security.

The market was not "wrong" in giving the security a price of 0.5 in equilibrium, given what was known then. It just happened to not work out for you as an investor, if it returned 0.

So with MBS. Given interest rates at issue, the market estimates a spread over the comparable T Bond. However that OAS will change over time.

If interest rates move by more than the market was expecting, then the market was not incorrect in its pricing, it's just more information has emerged.

Repayment and refinancing are interesting, because they are not just driven by volatility of interest rates, up or down. Things like making it easy to refinance online came in in the 1990s and increases the percentage of mortgage borrowers who would refinance for any given interest rate change.

You posed a question, and it's got a least 3 possible answers (which could add to 100%, or one could be 120% one way and the other -20% the other and the third 0%):

- in fact the risks are not very great. Negative convexity just does not show up to any great extent

- we just haven't had the interest rate conditions which would cause a big discrepancy between US Treasury bonds and GNMAs

- a third factor which has been true over that period, of falling interest rates, has been more important in setting the prices of US government securities

Both Larry Swedroe's book and David Swensen's take you through why investments in US govt MBS securities are not a good idea for individual investors.

*negative convexity is the 2nd derivative, so it's only for a given Price-Yield combination, for an infinitesimal (small) change that it measures that

Valuethinker
Posts: 36314
Joined: Fri May 11, 2007 11:07 am

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by Valuethinker » Sun Feb 04, 2018 12:03 pm

There's a couple of lines of argument going on here:

- (invalid argument) GNMA bonds have not performed differently than US Treasury bonds historically, therefore there is no such thing as extension/ prepayment risk in Mortgage Backed Securities

(extension to that argument) markets clearly don't think it's a threat, because after all the bonds have returned about the same.

- (valid argument, but maybe incomplete) as above therefore that risk may not be the one that really matters to investors, other factors are more important in determining returns (and volatility) for investors

(the problem with 2 is we are in this era of an unprecedented fall in interest rates (since 1981) and that, besides credit risk, is what really drives bond prices).

In other words, one cannot deny a known property of US residential mortgages and the associated MBS. BTW AFAIK it's not a risk with MBS of any other country, due to different structures of mortgage market (where borrowing rates are either floating rate, or fixed typically for terms of up to 5 years).

User avatar
Doc
Posts: 8659
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by Doc » Sun Feb 04, 2018 1:53 pm

In response to Livesoft
Valuethinker wrote:
Sat Feb 03, 2018 2:12 pm
livesoft wrote:
Sat Feb 03, 2018 11:20 am
Blueskies123 wrote:
Sat Feb 03, 2018 11:14 am
I also noticed there is very little discussion about GNMAs. ...
On bogleheads.org various fads come and go. There has been in the past significant discussion about GNMAs, but it is true there hasn't been much, if any, discussion lately.

I consider GNMAs government-backed and not MBS. I consider MBS something backed by a place like the ol' Countrywide folks.
You are kidding, right? You understand the difference?

A Mortgage Backed Security is an MBS if the underlying asset pools is made up of mortgages, regardless of who issues it. It is the securitization process that makes it an MBS.

It so happens the US Treasury guarantees the GNMAs. A Canadian MBS is normally guaranteed by the Canadian Mortgage and Housing Corporation (CMHC).

However you still have the issues of US mortgages being repayable ahead of maturity, or outside the lock in period (typically 5 years in other countries, when you can then refinance, but you cannot do so, except with big penalties, before the fixed rate period expires).

Thus you still have prepayment risk or extension risk, which gives rise to the property of negative convexity in the bonds.
The negative convexity of MBS is what was Swedroe's main objection and it was made well before Lehman in his Bond book on 2006 pgs 254-264.

Disclosure

1) I avoid MBS (including GNMA) and also long bonds so I don't use a TBM type fixed income portfolio.
2) I keep Larry's Bond Book within reach of my computer. :sharebeer
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
munemaker
Posts: 3503
Joined: Sat Jan 18, 2014 6:14 pm

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by munemaker » Sun Feb 04, 2018 3:50 pm

Doc wrote:
Sun Feb 04, 2018 1:53 pm


The negative convexity of MBS is what was Swedroe's main objection and it was made well before Lehman in his Bond book on 2006 pgs 254-264.
As far as I can tell, this negative convexity has never adversely affected performance of the Vanguard GNMA fund, so is this just theoretical or what?

User avatar
Doc
Posts: 8659
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by Doc » Sun Feb 04, 2018 4:08 pm

munemaker wrote:
Sun Feb 04, 2018 3:50 pm
Doc wrote:
Sun Feb 04, 2018 1:53 pm


The negative convexity of MBS is what was Swedroe's main objection and it was made well before Lehman in his Bond book on 2006 pgs 254-264.
As far as I can tell, this negative convexity has never adversely affected performance of the Vanguard GNMA fund, so is this just theoretical or what?
Never? Cf Vangaurd Intermediate Term Treasury

Image

We may have a different definition of "never".

Don't get misled by fixed income data since the Fed dropped the Federal Funds Rate to "zero" after Lehman.

The point of the negative convexity is that at times while it may go in the "right" direction for the borrower, it goes in the wrong direction for the lender. As far as the GNMA fund is concerned you and I are the lender.
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.

jalbert
Posts: 3806
Joined: Fri Apr 10, 2015 12:29 am

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by jalbert » Sun Feb 04, 2018 5:22 pm

Just keep in mind that Vanguard's GNMA fund is actively managed and I'm pretty sure has outperformed historically (certainly compare to the GNMA ETF the last 6 years). It's maybe not the fairest way to assess relative performance unless you think the fund alpha is persistent.
Managing GNMA portfolios is one area where active management has beaten indexing with good reliability. These use fairly complex analytics to forecast which passthroughs are least susceptible to prepayments. And managing maturity and other techniques are used to manage extension risk. A GNMA with mortgages having 10 years left to maturity has very different convexity properties than a newly parcel of mortgages that are at the beginning of their 30-year term

Correlation
between an actively managed GNMA fund and a GNMA index fund is generally fairly low. You can also see that the actively managed fund had lower volatility and a higher return.

Holding fixed-rate mortgages is tough for a bank because they originate them so they all are acquired at original issue, and they use adjustable-rate leverage (deposit accounts) so rising rates are a double whammy.
Risk is not a guarantor of return.

stlutz
Posts: 4780
Joined: Fri Jan 02, 2009 1:08 am

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by stlutz » Sun Feb 04, 2018 5:32 pm

Correlation between an actively managed GNMA fund and a GNMA index fund is generally fairly low. You can also see that the actively managed fund had lower volatility and a higher return.
On the "portfolio and management" tab for the Vanguard fund, they report an r-squared vs. the index of .97 and a beta of 1.08, which is quite different from what you are saying (?):

https://personal.vanguard.com/us/funds/ ... true#tab=2

User avatar
patrick013
Posts: 2403
Joined: Mon Jul 13, 2015 7:49 pm

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by patrick013 » Sun Feb 04, 2018 5:33 pm

Image

It will be interesting to see what the chart looks like a year, and 2 years from now.
age in bonds, buy-and-hold, 10 year business cycle

NibbanaBanana
Posts: 224
Joined: Sun Jan 22, 2017 10:34 pm

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by NibbanaBanana » Sun Feb 04, 2018 7:46 pm

Valuethinker wrote:
Sun Feb 04, 2018 11:45 am

You posed a question, and it's got a least 3 possible answers (which could add to 100%, or one could be 120% one way and the other -20% the other and the third 0%):
Okay. So what are they for Total Bond?

jalbert
Posts: 3806
Joined: Fri Apr 10, 2015 12:29 am

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by jalbert » Sun Feb 04, 2018 11:29 pm

stlutz wrote:
Sun Feb 04, 2018 5:32 pm
Correlation between an actively managed GNMA fund and a GNMA index fund is generally fairly low. You can also see that the actively managed fund had lower volatility and a higher return.
On the "portfolio and management" tab for the Vanguard fund, they report an r-squared vs. the index of .97 and a beta of 1.08, which is quite different from what you are saying (?):

https://personal.vanguard.com/us/funds/ ... true#tab=2
Does Vanguard report the time interval for the beta and R^2 sample stats they computed? With portfoliovisualizer sample correlation of 60-day rolling daily returns since 3/1/2012 shows 0.33. Monthly return sample correlation is 0.79, and annual return sample correlation (of just 6 years) is 0.99.

The GNMA index ETF is just 6 years old, The difference could be explained by Vanguard’s R^2 covering a longer sample period, but, in the absence of information to the contrary, I think it is more likely that Vanguard is using annual returns.
Risk is not a guarantor of return.

jalbert
Posts: 3806
Joined: Fri Apr 10, 2015 12:29 am

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by jalbert » Sun Feb 04, 2018 11:48 pm

The mathematics of managing portfolios of of MBS, the tools, were not as sophisticated then. It was Lew Ranieri at Salomon who drove this market in the late 70s/ early 80s
In the 70’s GNMAs were a new asset class and even if the techniques were available there were not the range of maturities available on the secondary market. Also in the 70’s most or all of the underlying mortgages in GNMA passthroughs were assumable mortgages that did not need to be paid off when a house was sold, so extension risk was more acute at that time.

The real question is whether the GNMA market is efficient enough for GNMA risks to be compensated adequately. If so, then they should improve risk-adjusted return of a bond portfolio once the risk-reducing power of diversification is incorporated.

I think where I would be concerned is in a bond portfolio held by a residential real estate investor as there may be some correlated risk with residential real estate values.
Risk is not a guarantor of return.

User avatar
patrick013
Posts: 2403
Joined: Mon Jul 13, 2015 7:49 pm

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by patrick013 » Mon Feb 05, 2018 3:13 pm

jalbert wrote:
Sun Feb 04, 2018 11:48 pm

The real question is whether the GNMA market is efficient enough for GNMA risks to be compensated adequately. If so, then they should improve risk-adjusted return of a bond portfolio once the risk-reducing power of diversification is incorporated.

I think where I would be concerned is in a bond portfolio held by a residential real estate investor as there may be some correlated risk with residential real estate values.
The below is a list of potential GNMA's to be issued in March 2018. There
are enough coupons available to secure a discount or premium price for
15 years (less after assumed life is calc'd). So held to maturity or sold at
a cap gain if one occurs the security has little risk IMO. Save fees and buy
when and if rates peak later this year or next. What other risk would there
be ?

MONTH CUSIP # MIP # POOL (COUPON RATE) TERM
MARCH 2018 36179TTB1 MA5046 JM 02.500 15
MARCH 2018 36179TTC9 MA5047 JM 03.000 15
MARCH 2018 36179TTD7 MA5048 JM 03.500 15
MARCH 2018 36179TTE5 MA5049 SF 02.000 15
MARCH 2018 36179TTF2 MA5050 SF 02.500 15
MARCH 2018 36179TTG0 MA5051 SF 03.000 15
MARCH 2018 36179TTH8 MA5052 SF 03.500 15
MARCH 2018 36179TTJ4 MA5053 SF 04.000 15
MARCH 2018 36179TTK1 MA5054 SF 04.500 15
MARCH 2018 36179TTL9 MA5055 SF 05.000 15
MARCH 2018 36179TTM7 MA5056 SF 05.500 15
MARCH 2018 36179TTN5 MA5057 SF 06.000 15
MARCH 2018 36179TTP0 MA5058 SF 06.500 15
MARCH 2018 36179TTQ8 MA5059 SF 07.000 15
MARCH 2018 36179TTR6 MA5060 SF 07.500 15
MARCH 2018 36179TTS4 MA5061 SF 08.000 15
MARCH 2018 36179TTT2 MA5062 SF 08.500 15
MARCH 2018 36179TU45 MA5103 SF 01.500 15
age in bonds, buy-and-hold, 10 year business cycle

User avatar
Electron
Posts: 1776
Joined: Sat Mar 10, 2007 8:46 pm

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by Electron » Mon Feb 05, 2018 3:47 pm

nisiprius wrote:
Sat Feb 03, 2018 5:45 pm
Despite Larry Swedroe's expertise and all of his explanations of the potential risks in GNMAs, they've always sounded like, at worst, "risk of slight underperformance," ego risk rather than serious financial risk. Whenever I stare at the growth charts I sure don't see any place in thirty-seven years where the alleged potential risk has ever showed up.
Those are my feelings exactly. However, I will note that the long term outperformance by VFIIX appears to have been in the initial years. If you check the Morningstar Chart tab or Performance tab, the comparison is quite a bit different over the 15, 10, 5, 3, and 1 year periods provided. The GNMA fund is ahead year-to-date probably as a result of lower duration.

As I recall Vanguard's Bond Index fund experienced a significant tracking error in one period which could impact these comparisons.

In regards to using the GNMA fund, I'd only add two other points. One would have less fixed income diversification if holding only mortgage securities. Secondly, any value added by active management may not be consistent from year to year and there may be changes in the portfolio managers.
Electron

User avatar
Doc
Posts: 8659
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by Doc » Mon Feb 05, 2018 4:39 pm

Electron wrote:
Mon Feb 05, 2018 3:47 pm
nisiprius wrote: ↑Sat Feb 03, 2018 4:45 pm
Despite Larry Swedroe's expertise and all of his explanations of the potential risks in GNMAs, they've always sounded like, at worst, "risk of slight underperformance," ego risk rather than serious financial risk. Whenever I stare at the growth charts I sure don't see any place in thirty-seven years where the alleged potential risk has ever showed up.
Those are my feelings exactly. However, I will note that the long term outperformance by VFIIX appears to have been in the initial years. If you check the Morningstar Chart tab or Performance tab, the comparison is quite a bit different over the 15, 10, 5, 3, and 1 year periods provided. The GNMA fund is ahead year-to-date probably as a result of lower duration.
Guys this comparison is cyclical. If you use growth charts or 15, 10, 5, 3, 1 periods you will miss the wiggles. You need to look at rolling return charts so you can see the trees in the forest.

See the two year rolling return chart for the last ~25 years that I posted earlier.

To make sure that we are on all on the same page is the question GNMAs vs. intermediate term Treasuries? Both have similar duration and Government guarantee.
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.

jalbert
Posts: 3806
Joined: Fri Apr 10, 2015 12:29 am

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by jalbert » Mon Feb 05, 2018 4:41 pm

patrick013 wrote:
Mon Feb 05, 2018 3:13 pm
jalbert wrote:
Sun Feb 04, 2018 11:48 pm

The real question is whether the GNMA market is efficient enough for GNMA risks to be compensated adequately. If so, then they should improve risk-adjusted return of a bond portfolio once the risk-reducing power of diversification is incorporated.

I think where I would be concerned is in a bond portfolio held by a residential real estate investor as there may be some correlated risk with residential real estate values.
The below is a list of potential GNMA's to be issued in March 2018. There
are enough coupons available to secure a discount or premium price for
15 years (less after assumed life is calc'd). So held to maturity or sold at
a cap gain if one occurs the security has little risk IMO. Save fees and buy
when and if rates peak later this year or next. What other risk would there
be ?

MONTH CUSIP # MIP # POOL (COUPON RATE) TERM
MARCH 2018 36179TTB1 MA5046 JM 02.500 15
MARCH 2018 36179TTC9 MA5047 JM 03.000 15
MARCH 2018 36179TTD7 MA5048 JM 03.500 15
MARCH 2018 36179TTE5 MA5049 SF 02.000 15
MARCH 2018 36179TTF2 MA5050 SF 02.500 15
MARCH 2018 36179TTG0 MA5051 SF 03.000 15
MARCH 2018 36179TTH8 MA5052 SF 03.500 15
MARCH 2018 36179TTJ4 MA5053 SF 04.000 15
MARCH 2018 36179TTK1 MA5054 SF 04.500 15
MARCH 2018 36179TTL9 MA5055 SF 05.000 15
MARCH 2018 36179TTM7 MA5056 SF 05.500 15
MARCH 2018 36179TTN5 MA5057 SF 06.000 15
MARCH 2018 36179TTP0 MA5058 SF 06.500 15
MARCH 2018 36179TTQ8 MA5059 SF 07.000 15
MARCH 2018 36179TTR6 MA5060 SF 07.500 15
MARCH 2018 36179TTS4 MA5061 SF 08.000 15
MARCH 2018 36179TTT2 MA5062 SF 08.500 15
MARCH 2018 36179TU45 MA5103 SF 01.500 15
Buying individual GNMAs is not a good idea for an individual investor. Prepayment and extension risk is enough higher at original issue that I believe institutional investors often if not usually are able to buy them below par at original issue. This would constitute a hidden subsidy of FHA and VA home buyers by the govt, implicitly buying down the mortgage rate for the home buyer a little compared to what the market will offer for the mortgage, on top of the risk underwriting that GNMA is mostly known for. A retail investor will almost certainly pay par plus commission, overpaying. And an individual GNMA may have liquidity problems since most demand is from institutional buyers looking for larger parcels. An individual investor also most likely does not have the analytical tools, expertise, or portfolio volume to manage risks and price prepayment options properly. These are complex investments best left to professional portfolio managers who specialize in them, and have the expertise to manage them properly.
Risk is not a guarantor of return.

User avatar
patrick013
Posts: 2403
Joined: Mon Jul 13, 2015 7:49 pm

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by patrick013 » Mon Feb 05, 2018 5:18 pm

jalbert wrote:
Mon Feb 05, 2018 4:41 pm
These are complex investments best left to professional portfolio managers who specialize in them, and have the expertise to manage them properly.
Alot of people buy them thru banks that have some kind of investment
brokerage service. The bank gets a concession from the underwriter.
They love the way the prepayments come in thus lowering principal
while still receiving the yield of a longer dated bond. Need to be happy
with the coupon, I always would wait till an issue is available at a discount
price, not much more to it than that. Those and Farm Credit Bank bonds
thru a local bank, but those are usually available at Fidelity, at the right price.
age in bonds, buy-and-hold, 10 year business cycle

User avatar
munemaker
Posts: 3503
Joined: Sat Jan 18, 2014 6:14 pm

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by munemaker » Mon Feb 05, 2018 5:29 pm

Doc wrote:
Sun Feb 04, 2018 4:08 pm
munemaker wrote:
Sun Feb 04, 2018 3:50 pm
Doc wrote:
Sun Feb 04, 2018 1:53 pm


The negative convexity of MBS is what was Swedroe's main objection and it was made well before Lehman in his Bond book on 2006 pgs 254-264.
As far as I can tell, this negative convexity has never adversely affected performance of the Vanguard GNMA fund, so is this just theoretical or what?
Never? Cf Vangaurd Intermediate Term Treasury

Image

We may have a different definition of "never".

Don't get misled by fixed income data since the Fed dropped the Federal Funds Rate to "zero" after Lehman.

The point of the negative convexity is that at times while it may go in the "right" direction for the borrower, it goes in the wrong direction for the lender. As far as the GNMA fund is concerned you and I are the lender.
I was comparing to vbtlx where the 5, 10 and 15 year returns are nearly identical. What difference does any of this make if the end result is the same yields? And if the undiversified asset (Vanguard GNMA fund) is government guaranteed, why do you need diversification? Just trying to understand the argument against the Vanguard GNMA fund, and I do not.

User avatar
Doc
Posts: 8659
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by Doc » Mon Feb 05, 2018 6:11 pm

munemaker wrote:
Mon Feb 05, 2018 5:29 pm
I was comparing to vbtlx where the 5, 10 and 15 year returns are nearly identical. What difference does any of this make if the end result is the same yields? And if the undiversified asset (Vanguard GNMA fund) is government guaranteed, why do you need diversification? Just trying to understand the argument against the Vanguard GNMA fund, and I do not.
Looking at the rolling return chart there are periods of 3 to 5 years where one leads and then the other leads. If you look at longer periods you don't see that because it averages out. The shorter term variation is at least partly do to the negative convexity. The lender, that's us the GNMA bond holders should be compensated for accepting the mortgage holders ability to prepay or extend their loan when rates move. Swedroe's point is that the compensation is inadequate. You can't get that information from looking at return comparisons with Treasuries over long time periods.

If you choose not to accept the bond experts idea that's your choice. I don't see any need to make that choice. I can just avoid MBS and get the same compensation with Treasuries.
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
patrick013
Posts: 2403
Joined: Mon Jul 13, 2015 7:49 pm

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by patrick013 » Mon Feb 05, 2018 6:35 pm

Doc wrote:
Mon Feb 05, 2018 6:11 pm
Swedroe's point is that the compensation is inadequate. You can't get that information from looking at return comparisons with Treasuries over long time periods.

If you choose not to accept the bond experts idea that's your choice. I don't see any need to make that choice. I can just avoid MBS and get the same compensation with Treasuries.
Doesn't that happen just for bonds previously at a premium ? I'm getting
the same yield on cost, same principal and any prepayment, thru the life
of the GNMA. From my perspective it's great for cash flow.
age in bonds, buy-and-hold, 10 year business cycle

User avatar
Doc
Posts: 8659
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by Doc » Mon Feb 05, 2018 7:11 pm

patrick013 wrote:
Mon Feb 05, 2018 6:35 pm
Doesn't that happen just for bonds previously at a premium ?
No. The US Treasury cannot say "Oh, interest rates have gone up. I'm going to delay paying off the loan." But the home owner can say, " Mortgage rates are high, I'm not going to buy my new house now because I can't 't afford the higher mortgage rate." The home owner wins. The GNMA holder loses. The question is does the GNMA holder get enough compensation for that option that is given to the home owner.

It also works the other way around. If rates go down the home owner refinances. And we, the GNMA holder get "hosed".
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: 4780
Joined: Fri Jan 02, 2009 1:08 am

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by stlutz » Mon Feb 05, 2018 8:41 pm

I think some people are coming to different conclusions based on the same set of facts mainly because there are a few different ways to look at selecting bonds:

Approach #1) There are experts who spend all day doing nothing but valuing bonds. I'm not one of them. As such, it makes no sense for me to participate in determining the relative value of bond X vs. bond Y. I'm going to buy the entire bond market. Find me the people who have in reality done much better that doing this with minimal costs. Not many of them actually exist.

Approach #2) Bonds are primarily about being a safe part of your portfolio and hopefully they will also provide at least some minimal positive return. You should not take risk in bonds unless you are well compensated for doing so. So, yes, you should participate in price setting of bond X vs. bond Y.

Approach #3) There are certain risks in bond investing that are absolutely never worth taking. Junk bonds, callable bonds etc. are all setup to be disadvantageous to the lender. One should always refuse to play that game. It you want to take big risks, invest in something that offers a big potential payoff (i.e. stocks) as opposed to something offering the possibility of some fraction of a percent of additional return.

The various "sides" in the discussion all seems to agree that including mortgage bonds in one's portfolio has neither substantially increased nor decreased returns. Those following approach #1 therefore conclude that there is no good reason to exclude such bonds from one's portfolio--they clearly aren't doing any harm. Those following approach #3 would say, "See, there's no benefit--so why are you taking the risk?" Those following approach #2 are saying that the answer of whether to include or exclude depends on price--history shows that this is not a set and forget kind of thing if you are looking for superior returns.

I personally am heavily overweight Treasuries and CDs relative to the market in my portfolio. I am persuaded by the backtesting showing that in a portfolio context, treasuries only does better.

However, there is a problem. Back in the late 90s the whole DFA crowd stressed that one should only use *short-term* Treasuries (avg. maturity of 1-2 years) as there has not been any historical benefit to taking term risk. What followed were two very large bear markets where the best kinds of bonds to own were long-term Treasuries. Moreover, post 2010 those in ST bonds have missed out on a lot of return. Oops.

While I might be persuaded of my own abilities to select the "best" fixed income securities for my portfolio, the historical evidence regarding backtesting suggests that those who are buying the entire market are actually smarter than those of us doing a bunch of analyzing.

User avatar
munemaker
Posts: 3503
Joined: Sat Jan 18, 2014 6:14 pm

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by munemaker » Tue Feb 06, 2018 8:03 am

Doc wrote:
Mon Feb 05, 2018 6:11 pm
munemaker wrote:
Mon Feb 05, 2018 5:29 pm
I was comparing to vbtlx where the 5, 10 and 15 year returns are nearly identical. What difference does any of this make if the end result is the same yields? And if the undiversified asset (Vanguard GNMA fund) is government guaranteed, why do you need diversification? Just trying to understand the argument against the Vanguard GNMA fund, and I do not.
Looking at the rolling return chart there are periods of 3 to 5 years where one leads and then the other leads. If you look at longer periods you don't see that because it averages out. The shorter term variation is at least partly do to the negative convexity. The lender, that's us the GNMA bond holders should be compensated for accepting the mortgage holders ability to prepay or extend their loan when rates move. Swedroe's point is that the compensation is inadequate. You can't get that information from looking at return comparisons with Treasuries over long time periods.

If you choose not to accept the bond experts idea that's your choice. I don't see any need to make that choice. I can just avoid MBS and get the same compensation with Treasuries.
Again, getting back to comparing the Vanguard GNMA fund to the Vanguard Total Return Fund, the short and long term returns are practically identical, but on one hand you have a mix of government and corporate bonds, where on the other hand you have 100% government guaranteed obligations. So why should I logically prefer the Total Bond Fund over the GNMA fund. More diversification has been stated. Diversification is for improving returns and reducing risk, and in this comparison, it does not seem to do either.

User avatar
Doc
Posts: 8659
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by Doc » Tue Feb 06, 2018 9:32 am

munemaker wrote:
Tue Feb 06, 2018 8:03 am
Again, getting back to comparing the Vanguard GNMA fund to the Vanguard Total Return Fund, the short and long term returns are practically identical, but on one hand you have a mix of government and corporate bonds, where on the other hand you have 100% government guaranteed obligations. So why should I logically prefer the Total Bond Fund over the GNMA fund. More diversification has been stated. Diversification is for improving returns and reducing risk, and in this comparison, it does not seem to do either.
I'm not comparing GNMA with Vg Total Return. That apples and oranges. Comparing GNMA with an Intermediate Treasury fund would be reasonable since they both have similar durations and are backed by the Government. That's what I did and that's what patrick013 did. My only point was that a total return chart or long term total return numbers hide information because the correlation between the two is cyclical and you can't see the cycles with a long term metric.

"Diversification is for improving returns and reducing risk." :o It is generally accepted that higher returns usually means higher risk. Think stocks and bonds. Investors expect to be rewarded for taking on more risk. To the extent that diversification reduces risk it would also be expected to reduce returns.
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
triceratop
Moderator
Posts: 5717
Joined: Tue Aug 04, 2015 8:20 pm
Location: la la land

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by triceratop » Tue Feb 06, 2018 9:43 am

Let me ask a slightly provacative question. There are a number of people who have stated that the market prices this risk and therefore it should improve your risk-adjusted return (I only hold IT treasuries so obviously I disagree, just for disclosure). So the question is: how is this risk priced? No seriously, what is the model for the bond and how do prevailing interest rates result in a given yield for GNMAs. We have to remember that the call options on these bonds are at least partially human behavior dependent; so the one paper out of Harvard Business I have seen (with math that makes me squeamish, and I do math for my day job) uses various trigger thresholds for how homeowners react to various changes in interest rates away from those at the time of their purchase. Show me how all this is priced and we can have a discussion that GNMAs have a role to play as a risk asset. Otherwise no. That is my provacative question. :)
"To play the stock market is to play musical chairs under the chord progression of a bid-ask spread."

User avatar
Doc
Posts: 8659
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by Doc » Tue Feb 06, 2018 9:47 am

stlutz wrote:
Mon Feb 05, 2018 8:41 pm
I think some people are coming to different conclusions based on the same set of facts mainly because there are a few different ways to look at selecting bonds:

...

While I might be persuaded of my own abilities to select the "best" fixed income securities for my portfolio, the historical evidence regarding backtesting suggests that those who are buying the entire market are actually smarter than those of us doing a bunch of analyzing.
I like your three factor idea but have trouble with the conclusion. If the bond segments don't matter and buying the total bond market has the same return then why not only buy those bonds segments that tend to have negative correlation with the equity market especially in times of stress.

I would suggest taking the Total Bond Market and throwing out those parts that don't work well with equities like GNMAs and the rest of the MBS along with other securitized "stuff". You would wind up with something like the Bloomberg Barclays Intermediate Government/Credit Index instead of the TBM. :idea:

Dividing the Intermediate Government/Credit Index into separated Treasury and Credit components might give you some benefit it an equity market crash but that's getting way beyond the scope of this thread.
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
munemaker
Posts: 3503
Joined: Sat Jan 18, 2014 6:14 pm

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by munemaker » Tue Feb 06, 2018 10:10 am

Doc wrote:
Tue Feb 06, 2018 9:32 am
munemaker wrote:
Tue Feb 06, 2018 8:03 am
Again, getting back to comparing the Vanguard GNMA fund to the Vanguard Total Return Fund, the short and long term returns are practically identical, but on one hand you have a mix of government and corporate bonds, where on the other hand you have 100% government guaranteed obligations. So why should I logically prefer the Total Bond Fund over the GNMA fund. More diversification has been stated. Diversification is for improving returns and reducing risk, and in this comparison, it does not seem to do either.
I'm not comparing GNMA with Vg Total Return. That apples and oranges. Comparing GNMA with an Intermediate Treasury fund would be reasonable since they both have similar durations and are backed by the Government. That's what I did and that's what patrick013 did. My only point was that a total return chart or long term total return numbers hide information because the correlation between the two is cyclical and you can't see the cycles with a long term metric.

"Diversification is for improving returns and reducing risk." :o It is generally accepted that higher returns usually means higher risk. Think stocks and bonds. Investors expect to be rewarded for taking on more risk. To the extent that diversification reduces risk it would also be expected to reduce returns.
We are talking two different things. I now realize you are not interested in addressing my question, but maybe someone else can.

If the returns of the VG Total Bond Market Fund are identical to the VG GNMA fund, why is the TBM fund preferable when it has higher risk and the same return?

User avatar
Doc
Posts: 8659
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by Doc » Tue Feb 06, 2018 10:49 am

munemaker wrote:
Tue Feb 06, 2018 10:10 am
If the returns of the VG Total Bond Market Fund are identical to the VG GNMA fund, why is the TBM fund preferable when it has higher risk and the same return?
You still have the correlation issue if you only look at total returns even if the Total Return is similar.

Image

But it is not as big a difference as the previous pairing.

But the GNMA fund did hold up somewhat better perhaps than TBM in the '08 crisis but not as well as an intermediate Treasury fund.

Image

You can choose whatever is best for your personal situation. For my situation I don't want either.
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.

garlandwhizzer
Posts: 2013
Joined: Fri Aug 06, 2010 3:42 pm

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by garlandwhizzer » Tue Feb 06, 2018 11:37 am

I think that arguing whether ITT or TBM or GNMA is the only one to own is akin to clerics arguing over how many angels could dance on a pinhead during the middle ages. All 3 are viable choices, so much less volatile than stocks and solid anchors in equity market declines. The differences in yield are not signifiant enough to make important impacts on overall portfolio return. It's a very slight yield/risk tradeoff. All three fulfill the major function of bonds, stability to offset equity volatility.

I believe Triceratops has an strongly equity dominated portfolio (80% or 90%) and it is a rational choice for 100% Treasuries in that situation which dollar for dollar offer maximal diversification to equity. If you don't have a large allocation to bonds you probably want maximal dollar for dollar diversification. That fits her circumstances, but others in different financial/portfolio circumstances may rationally choose other options. Personally in addition to TBM I also hold IT and ST Investment Grade Corporates, ST TIPS, and Prime MM, trying to cover all bases. Relative to ITT this raises the default risk a bit, lowers duration risk a lot, lowers unexpected inflation risk a bit, raises yield return a bit, and has a modest degree of equity volatility--all of which are fine with me and fit my circumstances. I am in the withdrawal phase and need to be able to sell bond assets periodically so I can choose to sell any of the above at any time depending market circumstances on an ongoing basis.

Garland Whizzer

User avatar
triceratop
Moderator
Posts: 5717
Joined: Tue Aug 04, 2015 8:20 pm
Location: la la land

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by triceratop » Tue Feb 06, 2018 11:55 am

Interesting reading, that I mentioned earlier: Mortgage Convexity - Samuel Hanson. Are others aware of better resources for how these securities are priced?

Oh, Garland, I happen to be male. :wink: I agree with your analysis. However, I would still like to leave my provocative question open.
"To play the stock market is to play musical chairs under the chord progression of a bid-ask spread."

User avatar
Doc
Posts: 8659
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: Why GNMAs Shouldn't Be Your Bond Choice

Post by Doc » Tue Feb 06, 2018 12:28 pm

triceratop wrote:
Tue Feb 06, 2018 11:55 am
However, I would still like to leave my provocative question open.
Larry E. Swedroe & Joseph H. Hempen wrote:What makes MBS <GNMA> so attractive to investors? The attraction is the higher stated yields they carry compared to other forms of fixed-income securities of comparable credit quality. The higher yields, and thus the greater expected <source emphasis> return come at the price, however, of greater risk. That risk is not in the form of credit risk, but, instead, takes the form of interest rate and duration (maturity) risk. The reason is that the maturity of an MBS is uncertion.
THE ONLY GUIDE TO A WINNING BOND STRATEGY YOU'LL EVER NEED" Copyright 2006
First Edition 2006 p156

For those of you who like to use growth charts and 5, 10, 15 year return numbers be aware that because of FED action since 2008 which kept interest rates low and relatively constant until just recently, that interest rate risk has not materialized so the metrics used are subject to a high degree of recency bias. (My opinion.)
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