Here's a statement of the obvious: The opinions expressed here are those of the participants, not those of the Mutual Fund Observer. We cannot vouch for the accuracy or appropriateness of any of it, though we do encourage civility and good humor.
So I guess my question is -- if interest rates rise, and provided the companies who have issued the bonds don't default or go bankrupt; the money that is lost is opportunity cost and inflation costs; correct (provided I don't have to sell my shares of a bond fund / bonds at a loss)?
This is the kind of article that makes people focus on duration and make bad choices as I mentioned in the other thread. Duration is most relevant in a bond fund that is the purest in the metric that is going up, so for example intermediate treasuries when those rates go up.
When bond funds have multiple sectors, it becomes very unreliable as an indicator since the velocity and volatility of the increase have a significant effect on different bond types to different extents.
For example, rapid increases will affect the economy to the extent that bonds that are correlated with economy like high yield will be significantly affected than a gradual increase. When rates increase to affect mortgage rates, the amount of refinancing decreases which leads to less mortgage loans being recalled which increases the value of existing mortgage bonds.
To the question raised by @shostakovich, bond funds have other side effects. Since these funds have to mark to market on a daily basis, their NAVs might dip first before the yield catches up. If this results in a lot of outflow of money from that fund, the fund may have to sell bonds to meet redemptions. If they sell the older lower yield bonds, they take a capital loss. If they sell the newer higher yield bonds, their yield drops down.
So, yes, if you have a pure Treasury fund, decrease the duration if you think rates will go up and take the potential opportunity cost loss if the rates do not go up as fast or as much as expected.
The irony is that while people are focused on bond funds in rate increases, a rapid rate increase will affect equities negatively much more than it will affect bonds in an overvalued equity market with larger losses.
If the rate increase is sudden from an overheating economy and/or inflation pressures of which there are very little signs, worry more about short term equity exposure. A declining equity market will support bond prices.
If it is a Fed orchestrated gradual increase with very little inflation pressures which is the likely scenario, don't worry about bonds and maintain a diversified portfolio in a good multi sector bond fund or spread across multiple funds each of which behave differently.
Since these funds have to mark to market on a daily basis, their NAVs might dip first before the yield catches up. If this results in a lot of outflow of money from that fund, the fund may have to sell bonds to meet redemptions. If they sell the older lower yield bonds, they take a capital loss. If they sell the newer higher yield bonds, their yield drops down.
I believe it depends upon the coupons and the type of yield (SEC, current yield, etc.) one is talking about.
For example, suppose you have two bonds from the same series/issuer (and equal face value) in the portfolio with a 3% coupon. Suppose current market rate for these bonds (given their credit rating, call/maturity date) is 3%. Suppose one was purchased a year ago (when rates were 2%), so it was purchased at a premium. It now trades at par because coupon matches market rate. The second bond was purchased yesterday at par.
First, note that just because a bond is older (to the portfolio) doesn't mean its coupon is less. Second, in this example at least, it doesn't matter to the yield which bond is sold, as they are identical bonds.
Because of mark to market, I believe the SEC yield would be unaffected even if these were not identical bonds - that is, all the bonds in the portfolio (of a given maturity/call date) should have the same YTW.
Current yield is a different story. If one bond is a discount bond (lower than market rate coupon, and priced lower than par), while a second bond is priced at par (coupon rate matching market rate), the former will have a lower current yield.
[Even though the former's price is a discount (so its current yield is higher than its coupon rate), its current yield is still below market. Otherwise, you'd get both market rate interest and appreciation to par, which is too much total return.]
So in this situation, selling the lower coupon bond would indeed increase current yield (though not SEC yield). I'm guessing that this is the effect you had in mind. However, as the example above shows, there's no reason to assume that it is the older bond that has the lower coupon.
@msf, I really don't understand what you are trying to get at because you have mixed up a couple of terms.
To standardize on terminology for the common case, consider bonds that pay interest on a regular basis than all at once at maturity and whose coupon rate is at market rate at issue so the price at issuance is the same as face vaue. The coupon rate by definition does not change (not considering floating rates, etc).
What is relevant to a future mutual fund buyer is the SEC yield. What is relevant to the bond manager and to the NAV is yield to maturity based on which the price of a bond in the secondary market is priced and the mark to market tries to estimate that.
Take a bond fund that buys only 10 year bonds as a simple example. Say it bought one such bond 5 years ago. The current price of that bond, assuming an efficient market, is similar to the price of another bond which has the same yield to maturity. So, for example, it may be similar to the price of a new issue of a 5 yr bond with similar credit risk. The current NAV reflects that.
The secondary market for bonds is not like equity markets. The liquidity is low and two bonds even from the same issuer separated by time may have different terms. So, it is more like buying real estate. You also have to find a buyer that wants the remaining duration bond. So, the spreads are higher. If any bond manager HAS to sell bonds, they usually take a loss. The mark to market does not capture this. It is more like the difference between an assessed value of a house vs the actual selling price when you have to sell it quickly.
If this manager buys another 10 year bond after 5 years, and the interest has gone up and the yield curve is not inverted, the SEC yield for the fund goes up at the time of purchase as does the effective duration of the fund which was 5 years before purchase. NAV doesn't change at that instant as cash is converted to face value. If the manager immediately sells the second bond, the SEC yield Fdrops to reflect the yield of the older bond with lower duration left and NAV drops by any loss incurred from spreads. The closer to issuance, typically the less of a capital loss. But the bond no longer gets the higher interest payout from the new bond.
If the manager sells the older bond instead, he will likely realize a larger capital loss from the spread that drops the NAV from last mark to market. The SEC yield for the fund goes up with the newer bond remaining but existing investors see the drop in NAV and panic and the outflow gets worse. This snowballs.
The situation is more complicated with a mix of bonds but no bond manager gains by HAVING to sell a bond in secondary market.
If you are pointing an exception or a special case, it is possible, but please do state the deviation from the above assumptions clearly without bringing in more variables than necessary to illustrate it.
Two identical bonds - same CUSIP, both bought on the secondary market. One, a year ago at a premium, the other yesterday at par. That was the example I was suggesting, and I think it has the minimum number of variables in it (as there aren't even two different coupons or maturity/call dates to consider).
This example contradicts a couple of statements, since the bonds are identical: - Selling the newer (more recently acquired) bond drops the fund's yield more than selling the earlier bond; and -The bond purchased earlier has a lower yield.
Bonds sold by the same issuer at different times are certainly different. Further, bonds sold by the same issuer at the same time are often different as well (different series). By sticking to identical bonds we eliminate all these differences in one fell swoop (or one swell foop).
Yes, there is a loss due to spread. But spread (and its effect) are likely close for two different bonds in the portfolio. That is, it's going to be problematic to specify conditions where the spread on one bond is necessarily greater than the spread on another similar, albeit not identical, bond.
Bond funds usually sell bonds on the secondary market (i.e. they do not hold bonds to maturity), so they have to eat the spread sooner or later. Still, I agree with you that being forced to sell sooner rather than later (increasing turnover) increases this cost.
I believe our communication problem is due to an assumption on your part that later-acquired bonds necessarily have later maturity/call dates than earlier-acquired bonds. In that case, as you stated, given the usual non-inverted yield curve, the later acquired bond will indeed have a higher YTW.
This (acquiring later and later maturing bonds as time passes) is a simplifying assumption, but I submit not necessarily accurate. Especially in the current environment where managers are methodically shortening duration.
"The closer to issuance, typically the less of a capital loss". Why? You wrote that when one buys a bond, the NAV doesn't change (except for loss due to spread). That is correct, and independent of the date of issue of the bond being purchased. No capital gain or loss on the purchase.
The reverse (opposite side of the trade) ought to be true as well - exchanging a bond for cash shouldn't affect the value of the portfolio (except for the trading costs, i.e. spread and commission). This is the point of marking to market - that the value of each bond used in computing the NAV is the FMV of the bond (excluding fire sales, as you observed, and excluding trading costs).
If by capital loss you meant the difference between the bond's sale price and its acquisition price, it seems that the closer to issuance (i.e. the further from maturity, assuming the same 10 year non-callable maturities) the greater the capital loss as interest rates rise.
Using the example you provided of a 10 year bond purchased five years ago and a 10 year bond purchased yesterday, the older bond has shorter duration. Thus:
- The older bond, with shorter duration, will see its price drop less as a consequence of an increase in interest rates; - The older bond will likely have a higher coupon (interest rates have been falling, so a bond issued five years ago was required to pay higher rates); the higher the coupon, the shorter the duration for a given maturity, so this further lessens the impact of rising rates.
Both of these argue for the older bond (further from issuance) dropping less in value, and thus realizing a lesser capital loss upon sale.
You may be mitigating this by assuming that interest rates went up between the time of purchase of the first bond (five years ago) and now so the older bond has an unrealized loss embedded. But since rates have generally declined over the past several years (and are still lower than they were five years ago), the older bond will typically have an unrealized gain, not loss.
We may need another round of exchanges to get everything clear. Probably not worth it, which is a point in itself - and one that you were making. That there are a lot of moving parts, and a lot of implicit assumptions. Which is why one cannot simply latch on to a single attribute (such as duration) and predict future results from that.
Feels like beating a dead horse. I still don't have any sense whether you are pointing to a very specific special case where the general doesn't hold or saying something more fundamental.
If the former, then fine. There us always a special case or an exception.
The ambiguity is inherent in the example you set up if buying the same bond earlier at a premium and later at par. What is the assumption behind this unusual scenario? Market inefficiencies? Interest rate changes? Manager stupidity? Is that a special case or the norm?
To simplify things, the manager is concerned about three things in his bond trading. YTM, duration and credit quality (do not bring in SEC yield, current yield, etc for this trading part, those are terms for the investors of the fund). For the same latter two, bonds will be priced to have the same YTM modulo market inefficiencies and trading friction.
In your example, the two purchases will be the same in all of those three whether he bought it earlier or later if you ignore inefficiencies and trading friction. So, yes, it makes no difference which one he sells in that scenario. If that is all you are saying as an exception, fine with me.
In a stable interest rate scenario, and ignoring credit quality variable, a manager will buy a lower YTM bond for a lower duration to reduce interest rate risk or buy the same YTM at similar duration to put cash to work. And yes there are exceptions of buying something cheap from a fire sale at PIMCO!
In a rising rate scenario, the manager can either get a higher YTM at the same duration or same YTM at a lower duration. So the prices for older issues with lower coupon rate will adjust downwards over time depending on liquidity to bring up the YTM to the same as newer bonds with same duration. This gets reflected in the NAV but you don't know the actual price until you sell it and this typically results in a further capital loss to meet the redemptions in addition to the NAV loss that has already happened in mark to market.
In terms of the dividends paid to remaining investors, the monthly dividends will actually go up if the newer buys maintained the duration or remain same or lower if they are lower in duration (think about this for a minute if you don't immediately see this without getting confused by current yields, SEC yields, etc). But the immediate drop in NAV may prompt further redemptions.
You can also sell the newer buys of bonds issued after the rate increase. My assumption is that these are much more efficiently marked to market until the rates rise again because the liquidity is higher with the recent coupon rates there being a lot more new issues at the prevalent coupon rates. But there may still be some capital loss because of spreads.
If the newer buys were lower duration buys then the duration of the fund increases (I think this is the case you are pointing to as an exception to my original post) but periodic dividends may remain the same and you can sell these. If they had the same duration as older buys, then the periodic dividends decrease after the sale.
By using duration and YTM, there is no assumption on the term to maturity of bonds and whether they were bought new or old in secondary markets. There is no confusion about premium, discount and par.
In both cases, the total return decreases. The choice for the manager is based on whether his investors are more sensitive to paper losses in NAV or reduction in monthly interest. On this we seem to agree, which was the original point.
Been occupied for awhile. Let me try this briefly, for I agree with you, we are not getting anywhere, and the horse is at least out of the barn door, if not completely dead.
The ambiguity is inherent in the example you set up if buying the same bond earlier at a premium and later at par. What is the assumption behind this unusual scenario? Market inefficiencies? Interest rate changes? Manager stupidity? Is that a special case or the norm?
Interest rate change - the assumption of the whole thread - that interest rates have gone up.
In your example, the two purchases will be the same in all of those three [YTM, duration, quality] whether he bought it earlier or later if you ignore inefficiencies and trading friction. So, yes, it makes no difference which one he sells in that scenario. If that is all you are saying as an exception, fine with me.
I reduced the number of variables to a minimum. If you feel that is oversimplified (thus exceptional), we can walk through other cases, though I prefer to recognize that the horse is dead. By focusing on one variable - rate changes and not bond differences, I felt we could look at pricing without getting bogged down in lots of other factors.
In a rising rate scenario, the manager can either get a higher YTM at the same duration or same YTM at a lower duration.
Here, I believe, is the source of confusion. In a rising rate scenario, the YTW of the bond in the portfolio rises in sync with the market. That's what mark to market does. The manager cannot buy a bond with a higher YTW than the bond in his portfolio sans market inefficiencies. That is, the NAV of the portfolio (the single bond, in this simplified discussion) has already dropped. And the portfolio yield has risen.
(Note that the dividends have not risen; the NAV has fallen. Which IMHO makes introducing dividends into the discussion more of a distraction than a clarification. I'm happy to work it through with you, though I doubt either of us feel it is worth the time/bandwidth/effort.)
So the prices for older issues with lower coupon rate will adjust downwards over time depending on liquidity to bring up the YTM to the same as newer bonds with same duration.
This gets reflected in the NAV but you don't know the actual price until you sell it and this typically results in a further capital loss to meet the redemptions in addition to the NAV loss that has already happened in mark to market.
Here you're talking about a distortion caused by a fire sale (inefficient market in that the seller is under duress to take a below-market price). I've already conceded that point, but question whether the impact is more substantial for one bond than another, especially when the two bonds share the same quality and duration.
Some of the comments you have following this seem to address how one bond's price is indeed distorted more than another. We can likewise discuss that (and it is an interesting consideration), but it goes beyond your original statement that was based on, as you wrote above, YTM, duration, and credit quality, not date of issue, maturity, etc.
The choice for the manager is based on whether his investors are more sensitive to paper losses in NAV or reduction in monthly interest. On this we seem to agree, which was the original point.
Yes. Unfortunately, investors are not always rational, forcing the fire sales and otherwise distorting the market to their detriment (and that of others who share in the same fund/investment).
There should not be any confusion. Interest rates will go up, but we do not know when or how fast. Likely scenario is no fed increase in Fed Funds Rate until late 2015, unless inflation moves faster than expected. When they do raise Fed Funds Rate, it will most likely be in quarter-percent increments. So what to do?
Don't spend a lot of time trying to out-think the Fed. If you own individual bonds, and you plan to hold until maturity, continue to hold. Or sell the bond, capture what is probably a sizeable gain, and move to flexible bond funds. If you own bond funds, use some logic. Avoid long-term Treasuries since they will suffer the worst fate. Avoid funds that can only use one kind of bond or a certain maturity range of bonds. Use funds where managers have a lot of flexibility. Duration itself may not be as important as duration compared to long-term Treasuries. Hence importance for flexibility of investment style. Look at funds like OSTIX, GSZIX, TSIIX, LSBDX.
@msf, I recommend that if you get a chance sit with a bond trader and see how bond trading is done. What I am saying will make a lot more sense then.
You are theoretically assuming things about Mark to Market that is close to equity markets than to the reality of bond markets which is more like real estate markets in price discovery.
Then see the whole post than a sequence of individual sentences. It will make much more sense.
These services practice a craft (not a science), just as art appraisers, tax assessors, etc. practice a craft. If the valuation (pricing for NAV) is on average fair (i.e. not biased to one side or the other), then on average the NAV will match the price fetched for actual sales (ignoring commissions, 1/2 spread, etc.)
It sounds like you're assuming that pricing services for bonds generally lag the market. For the sake of argument, let's assume this is true. (Heartland, circa 2000 comes to mind, but I digress.)
That leads to the idea that the older the last trade of a bond, the greater the lag (discrepancy between valuation and market price) on average. You seem to take this a step further, though.
You write that selling a less recently acquired bond (let's call it bond A) will suffer a capital loss in a falling market (i.e. will sell for less than its valuation), while a more recently acquired bond (let's call it Bond B) will not, or at least will suffer a lesser cap loss.
Here it seems you're making an unsubstantiated leap. Namely that bond B is the more recently traded bond. I gave the simplest counterexample - where bond A and bond B were identical. The accuracy of the valuation on each would have to be identical as well.
Perhaps my example stripped out too much (thus obscuring rather than informing). Take instead an illiquid bond B and a liquid, actively traded bond A. It is quite likely (though not certain) that bond A was traded on the market more recently than bond B, though bond A was traded by the fund less recently than bond B.
Here, the cap loss on bond B would be greater.
The accuracy of the bonds' valuation (and thus cap loss) depends on factors largely independent of the fund's specific trades. How liquid the bond is, how recently it traded on the market, depth of book, lot size, volume, etc. None of these are significantly affected by which bond the particular fund traded first.
(The cap loss due to lag would also seem to be affected by the duration of the bond, since prices move less on shorter duration bonds. I previously addressed the assumption that bond B has a longer duration - managers are shortening up on duration, so it is not clear that bond B's duration is later than bond A's.)
Just as you see me making, shall we say, unwarranted assumptions about marking to market, I respectfully suggest that you are also making some questionable assumptions about the trades that bond fund managers make, especially in the current market.
These services practice a craft (not a science), just as art appraisers, tax assessors, etc. practice a craft. If the valuation (pricing for NAV) is on average fair (i.e. not biased to one side or the other), then on average the NAV will match the price fetched for actual sales (ignoring commissions, 1/2 spread, etc.)
The above makes absolutely no sense whatsoever unless you have a very good price discovery mechanism.
I am not sure you are aware how bond trading works at all. This is like saying that on average the appraisal value if a house matches the sale price of a house. Not even close. This is because real estate market does not have a good price discovery mechanism. You can make an estimate based on comps. It will be in the ball park but often significantly different based on the motivation of the seller, or the buyer or the market condition, etc.
This is very different from the pricing of equities where thousand if not millions of shares of an equity can be traded in a day and there is a bid and ask on each equity. This is not what happens with bonds. I have a feeling that you are assuming that is how bond trading works with really no knowledge of bond markets.
It sounds like you're assuming that pricing services for bonds generally lag the market.
No. I am saying the appraised value is not a good indicator of what happens when you actually try to sell a bond. Like a house. In practice, not in some text book.
Even the fair value pricing of international equities which has much better pricing discovery is subject to significant deviations from where the market price turns out to be and often leads to significant corrections in the next day's pricing. Look at two comparable funds with similar portfolios and you will see them value differently on the same day when markets move. Fairness doesn't imply correctness. It just implies that it is not biased towards any one side. You are missing that distinction.
I am bowing out of this discussion with this because the difference between us is not assumptions on bond pricing but difference in understanding of how bond trading works. I don't feel it is possible to communicate meaningfully with that difference.
Comments
When bond funds have multiple sectors, it becomes very unreliable as an indicator since the velocity and volatility of the increase have a significant effect on different bond types to different extents.
For example, rapid increases will affect the economy to the extent that bonds that are correlated with economy like high yield will be significantly affected than a gradual increase. When rates increase to affect mortgage rates, the amount of refinancing decreases which leads to less mortgage loans being recalled which increases the value of existing mortgage bonds.
To the question raised by @shostakovich, bond funds have other side effects. Since these funds have to mark to market on a daily basis, their NAVs might dip first before the yield catches up. If this results in a lot of outflow of money from that fund, the fund may have to sell bonds to meet redemptions. If they sell the older lower yield bonds, they take a capital loss. If they sell the newer higher yield bonds, their yield drops down.
So, yes, if you have a pure Treasury fund, decrease the duration if you think rates will go up and take the potential opportunity cost loss if the rates do not go up as fast or as much as expected.
The irony is that while people are focused on bond funds in rate increases, a rapid rate increase will affect equities negatively much more than it will affect bonds in an overvalued equity market with larger losses.
If the rate increase is sudden from an overheating economy and/or inflation pressures of which there are very little signs, worry more about short term equity exposure. A declining equity market will support bond prices.
If it is a Fed orchestrated gradual increase with very little inflation pressures which is the likely scenario, don't worry about bonds and maintain a diversified portfolio in a good multi sector bond fund or spread across multiple funds each of which behave differently.
For example, suppose you have two bonds from the same series/issuer (and equal face value) in the portfolio with a 3% coupon. Suppose current market rate for these bonds (given their credit rating, call/maturity date) is 3%. Suppose one was purchased a year ago (when rates were 2%), so it was purchased at a premium. It now trades at par because coupon matches market rate. The second bond was purchased yesterday at par.
First, note that just because a bond is older (to the portfolio) doesn't mean its coupon is less. Second, in this example at least, it doesn't matter to the yield which bond is sold, as they are identical bonds.
Because of mark to market, I believe the SEC yield would be unaffected even if these were not identical bonds - that is, all the bonds in the portfolio (of a given maturity/call date) should have the same YTW.
Current yield is a different story. If one bond is a discount bond (lower than market rate coupon, and priced lower than par), while a second bond is priced at par (coupon rate matching market rate), the former will have a lower current yield.
[Even though the former's price is a discount (so its current yield is higher than its coupon rate), its current yield is still below market. Otherwise, you'd get both market rate interest and appreciation to par, which is too much total return.]
So in this situation, selling the lower coupon bond would indeed increase current yield (though not SEC yield). I'm guessing that this is the effect you had in mind. However, as the example above shows, there's no reason to assume that it is the older bond that has the lower coupon.
To standardize on terminology for the common case, consider bonds that pay interest on a regular basis than all at once at maturity and whose coupon rate is at market rate at issue so the price at issuance is the same as face vaue. The coupon rate by definition does not change (not considering floating rates, etc).
What is relevant to a future mutual fund buyer is the SEC yield. What is relevant to the bond manager and to the NAV is yield to maturity based on which the price of a bond in the secondary market is priced and the mark to market tries to estimate that.
Take a bond fund that buys only 10 year bonds as a simple example. Say it bought one such bond 5 years ago. The current price of that bond, assuming an efficient market, is similar to the price of another bond which has the same yield to maturity. So, for example, it may be similar to the price of a new issue of a 5 yr bond with similar credit risk. The current NAV reflects that.
The secondary market for bonds is not like equity markets. The liquidity is low and two bonds even from the same issuer separated by time may have different terms. So, it is more like buying real estate. You also have to find a buyer that wants the remaining duration bond. So, the spreads are higher. If any bond manager HAS to sell bonds, they usually take a loss. The mark to market does not capture this. It is more like the difference between an assessed value of a house vs the actual selling price when you have to sell it quickly.
If this manager buys another 10 year bond after 5 years, and the interest has gone up and the yield curve is not inverted, the SEC yield for the fund goes up at the time of purchase as does the effective duration of the fund which was 5 years before purchase. NAV doesn't change at that instant as cash is converted to face value. If the manager immediately sells the second bond, the SEC yield Fdrops to reflect the yield of the older bond with lower duration left and NAV drops by any loss incurred from spreads. The closer to issuance, typically the less of a capital loss. But the bond no longer gets the higher interest payout from the new bond.
If the manager sells the older bond instead, he will likely realize a larger capital loss from the spread that drops the NAV from last mark to market. The SEC yield for the fund goes up with the newer bond remaining but existing investors see the drop in NAV and panic and the outflow gets worse. This snowballs.
The situation is more complicated with a mix of bonds but no bond manager gains by HAVING to sell a bond in secondary market.
If you are pointing an exception or a special case, it is possible, but please do state the deviation from the above assumptions clearly without bringing in more variables than necessary to illustrate it.
This example contradicts a couple of statements, since the bonds are identical:
- Selling the newer (more recently acquired) bond drops the fund's yield more than selling the earlier bond; and
-The bond purchased earlier has a lower yield.
Bonds sold by the same issuer at different times are certainly different. Further, bonds sold by the same issuer at the same time are often different as well (different series). By sticking to identical bonds we eliminate all these differences in one fell swoop (or one swell foop).
Yes, there is a loss due to spread. But spread (and its effect) are likely close for two different bonds in the portfolio. That is, it's going to be problematic to specify conditions where the spread on one bond is necessarily greater than the spread on another similar, albeit not identical, bond.
Bond funds usually sell bonds on the secondary market (i.e. they do not hold bonds to maturity), so they have to eat the spread sooner or later. Still, I agree with you that being forced to sell sooner rather than later (increasing turnover) increases this cost.
I believe our communication problem is due to an assumption on your part that later-acquired bonds necessarily have later maturity/call dates than earlier-acquired bonds. In that case, as you stated, given the usual non-inverted yield curve, the later acquired bond will indeed have a higher YTW.
This (acquiring later and later maturing bonds as time passes) is a simplifying assumption, but I submit not necessarily accurate. Especially in the current environment where managers are methodically shortening duration.
"The closer to issuance, typically the less of a capital loss". Why? You wrote that when one buys a bond, the NAV doesn't change (except for loss due to spread). That is correct, and independent of the date of issue of the bond being purchased. No capital gain or loss on the purchase.
The reverse (opposite side of the trade) ought to be true as well - exchanging a bond for cash shouldn't affect the value of the portfolio (except for the trading costs, i.e. spread and commission). This is the point of marking to market - that the value of each bond used in computing the NAV is the FMV of the bond (excluding fire sales, as you observed, and excluding trading costs).
If by capital loss you meant the difference between the bond's sale price and its acquisition price, it seems that the closer to issuance (i.e. the further from maturity, assuming the same 10 year non-callable maturities) the greater the capital loss as interest rates rise.
Using the example you provided of a 10 year bond purchased five years ago and a 10 year bond purchased yesterday, the older bond has shorter duration. Thus:
- The older bond, with shorter duration, will see its price drop less as a consequence of an increase in interest rates;
- The older bond will likely have a higher coupon (interest rates have been falling, so a bond issued five years ago was required to pay higher rates); the higher the coupon, the shorter the duration for a given maturity, so this further lessens the impact of rising rates.
Both of these argue for the older bond (further from issuance) dropping less in value, and thus realizing a lesser capital loss upon sale.
You may be mitigating this by assuming that interest rates went up between the time of purchase of the first bond (five years ago) and now so the older bond has an unrealized loss embedded. But since rates have generally declined over the past several years (and are still lower than they were five years ago), the older bond will typically have an unrealized gain, not loss.
We may need another round of exchanges to get everything clear. Probably not worth it, which is a point in itself - and one that you were making. That there are a lot of moving parts, and a lot of implicit assumptions. Which is why one cannot simply latch on to a single attribute (such as duration) and predict future results from that.
If the former, then fine. There us always a special case or an exception.
The ambiguity is inherent in the example you set up if buying the same bond earlier at a premium and later at par. What is the assumption behind this unusual scenario? Market inefficiencies? Interest rate changes? Manager stupidity? Is that a special case or the norm?
To simplify things, the manager is concerned about three things in his bond trading. YTM, duration and credit quality (do not bring in SEC yield, current yield, etc for this trading part, those are terms for the investors of the fund). For the same latter two, bonds will be priced to have the same YTM modulo market inefficiencies and trading friction.
In your example, the two purchases will be the same in all of those three whether he bought it earlier or later if you ignore inefficiencies and trading friction. So, yes, it makes no difference which one he sells in that scenario. If that is all you are saying as an exception, fine with me.
In a stable interest rate scenario, and ignoring credit quality variable, a manager will buy a lower YTM bond for a lower duration to reduce interest rate risk or buy the same YTM at similar duration to put cash to work. And yes there are exceptions of buying something cheap from a fire sale at PIMCO!
In a rising rate scenario, the manager can either get a higher YTM at the same duration or same YTM at a lower duration. So the prices for older issues with lower coupon rate will adjust downwards over time depending on liquidity to bring up the YTM to the same as newer bonds with same duration. This gets reflected in the NAV but you don't know the actual price until you sell it and this typically results in a further capital loss to meet the redemptions in addition to the NAV loss that has already happened in mark to market.
In terms of the dividends paid to remaining investors, the monthly dividends will actually go up if the newer buys maintained the duration or remain same or lower if they are lower in duration (think about this for a minute if you don't immediately see this without getting confused by current yields, SEC yields, etc). But the immediate drop in NAV may prompt further redemptions.
You can also sell the newer buys of bonds issued after the rate increase. My assumption is that these are much more efficiently marked to market until the rates rise again because the liquidity is higher with the recent coupon rates there being a lot more new issues at the prevalent coupon rates. But there may still be some capital loss because of spreads.
If the newer buys were lower duration buys then the duration of the fund increases (I think this is the case you are pointing to as an exception to my original post) but periodic dividends may remain the same and you can sell these. If they had the same duration as older buys, then the periodic dividends decrease after the sale.
By using duration and YTM, there is no assumption on the term to maturity of bonds and whether they were bought new or old in secondary markets. There is no confusion about premium, discount and par.
In both cases, the total return decreases. The choice for the manager is based on whether his investors are more sensitive to paper losses in NAV or reduction in monthly interest. On this we seem to agree, which was the original point.
(Note that the dividends have not risen; the NAV has fallen. Which IMHO makes introducing dividends into the discussion more of a distraction than a clarification. I'm happy to work it through with you, though I doubt either of us feel it is worth the time/bandwidth/effort.) Exactly. Here you're talking about a distortion caused by a fire sale (inefficient market in that the seller is under duress to take a below-market price). I've already conceded that point, but question whether the impact is more substantial for one bond than another, especially when the two bonds share the same quality and duration.
Some of the comments you have following this seem to address how one bond's price is indeed distorted more than another. We can likewise discuss that (and it is an interesting consideration), but it goes beyond your original statement that was based on, as you wrote above, YTM, duration, and credit quality, not date of issue, maturity, etc. Yes. Unfortunately, investors are not always rational, forcing the fire sales and otherwise distorting the market to their detriment (and that of others who share in the same fund/investment).
Don't spend a lot of time trying to out-think the Fed. If you own individual bonds, and you plan to hold until maturity, continue to hold. Or sell the bond, capture what is probably a sizeable gain, and move to flexible bond funds. If you own bond funds, use some logic. Avoid long-term Treasuries since they will suffer the worst fate. Avoid funds that can only use one kind of bond or a certain maturity range of bonds. Use funds where managers have a lot of flexibility. Duration itself may not be as important as duration compared to long-term Treasuries. Hence importance for flexibility of investment style. Look at funds like OSTIX, GSZIX, TSIIX, LSBDX.
You are theoretically assuming things about Mark to Market that is close to equity markets than to the reality of bond markets which is more like real estate markets in price discovery.
Then see the whole post than a sequence of individual sentences. It will make much more sense.
Funds typically use a pricing service to price their securities, whether equities, bonds, derivatives, or other securities. http://www.icifactbook.org/2006/06_fb_appa.html
These services practice a craft (not a science), just as art appraisers, tax assessors, etc. practice a craft. If the valuation (pricing for NAV) is on average fair (i.e. not biased to one side or the other), then on average the NAV will match the price fetched for actual sales (ignoring commissions, 1/2 spread, etc.)
It sounds like you're assuming that pricing services for bonds generally lag the market. For the sake of argument, let's assume this is true. (Heartland, circa 2000 comes to mind, but I digress.)
That leads to the idea that the older the last trade of a bond, the greater the lag (discrepancy between valuation and market price) on average. You seem to take this a step further, though.
You write that selling a less recently acquired bond (let's call it bond A) will suffer a capital loss in a falling market (i.e. will sell for less than its valuation), while a more recently acquired bond (let's call it Bond B) will not, or at least will suffer a lesser cap loss.
Here it seems you're making an unsubstantiated leap. Namely that bond B is the more recently traded bond. I gave the simplest counterexample - where bond A and bond B were identical. The accuracy of the valuation on each would have to be identical as well.
Perhaps my example stripped out too much (thus obscuring rather than informing). Take instead an illiquid bond B and a liquid, actively traded bond A. It is quite likely (though not certain) that bond A was traded on the market more recently than bond B, though bond A was traded by the fund less recently than bond B.
Here, the cap loss on bond B would be greater.
The accuracy of the bonds' valuation (and thus cap loss) depends on factors largely independent of the fund's specific trades. How liquid the bond is, how recently it traded on the market, depth of book, lot size, volume, etc. None of these are significantly affected by which bond the particular fund traded first.
(The cap loss due to lag would also seem to be affected by the duration of the bond, since prices move less on shorter duration bonds. I previously addressed the assumption that bond B has a longer duration - managers are shortening up on duration, so it is not clear that bond B's duration is later than bond A's.)
Just as you see me making, shall we say, unwarranted assumptions about marking to market, I respectfully suggest that you are also making some questionable assumptions about the trades that bond fund managers make, especially in the current market.
I am not sure you are aware how bond trading works at all. This is like saying that on average the appraisal value if a house matches the sale price of a house. Not even close. This is because real estate market does not have a good price discovery mechanism. You can make an estimate based on comps. It will be in the ball park but often significantly different based on the motivation of the seller, or the buyer or the market condition, etc.
This is very different from the pricing of equities where thousand if not millions of shares of an equity can be traded in a day and there is a bid and ask on each equity. This is not what happens with bonds. I have a feeling that you are assuming that is how bond trading works with really no knowledge of bond markets. No. I am saying the appraised value is not a good indicator of what happens when you actually try to sell a bond. Like a house. In practice, not in some text book.
Even the fair value pricing of international equities which has much better pricing discovery is subject to significant deviations from where the market price turns out to be and often leads to significant corrections in the next day's pricing. Look at two comparable funds with similar portfolios and you will see them value differently on the same day when markets move. Fairness doesn't imply correctness. It just implies that it is not biased towards any one side. You are missing that distinction.
I am bowing out of this discussion with this because the difference between us is not assumptions on bond pricing but difference in understanding of how bond trading works. I don't feel it is possible to communicate meaningfully with that difference.