Yes, at present, TLT remains every bit as effective as it has been at any point throughout our data sample. My concerns are going forward into a new and vastly different regime. That applies to return, volatility and correlations.
I think there are really two ways TLT is being used by people here.
One is as a more-or-less fixed or at least stable part of the portfolio, some sort of SPY/TLT or stocks/TLT mixture. That, I think, is likely to be an unmitigated disaster going forward for reasons explained in the Forbes posting.
The other is as a short-term hedge – something constructive to use as an alternative to stocks when timing models turn bearish on the latter.
That’s a different and interesting question. Essentially, the idea is that regardless of big-picture concerns, when cash gets nervous and needs to go or park somewhere quickly, TLT can be of value. In a big picture sense, we might look back upon it as a temporary upward phase n the context of a larger bear market – these things happen with all assets, so there’s no reason to assume TLT would be an exception.
Given that no asset trends in a straight line in any direction and that short-term movements of money can cause anything at any time, then yes, I would have t pull back on my previous assertions that TLT will be completely ineffective. But that still leaves the question of whether it’s as effective as other hedge choices.
Cash is a hedge choice (in our sims, it provides zero return but in real life, the return is alway positive, and in a period of rising interest rates that will become increasingly positive, and its volatility is, for all practical purposes, zero. So that means it always deserves at least a place at the table when one is choosing among possible hedges.
SHY is another. It’s TLT like in many respect, most particularly the safe harbor role it can play as cash shits quickly. But being the short-term version of TLT, it is inherently much less volatile, meaning that it didn’t make the cut in sims which, in a bond bull market, had no choice but to favor TLT could beat TLT soundly in a period of rising rates.
SH is another. If market timing hedging is the goal, there’s something to be said for shorting equities (and SH allows one to do so without the challenges that come from the leveraged versions of short ETFs). What’s interesting here is the built-in -1.00 correlation to SPY (subject to tracking error). It means we only have to consider and assess the fundamentals and technicals of equities. We’re spared the burden of trying to guess how one tiny piece of the economic pie (interest rates) will play out in the yield curve. We can work well with what we have and what we know best.
There’s GLD. Darned if I understand how gold moves. But it’s a fun vehicle and how could I talk abut potential hedge choices without at least paying homage.
And then, there’s TLT, the perennial winner in just about all quantitative studies done on our platform. Going forward, as I backed up and acknowledged, yes, it could provide the sort of counter-trend blips that would call to mind its good old days. So if your timing model is really precise, then yes, TLT has much going for it. It is the most volatile way we can play fixed income (due to it’s built-in 20+ year portfolio) so yes, if the market falls hard on any given day or span of days, TLT may again rise most. But that changes the challenge. The thing that makes tLT attractive if your timing is spot on also makes it deadly if you miss the mark. Because TLT is inherently volatile, and because any gains it makes going forward will be counter-trend bear-market rallies (unlike in the past, when the gains were mainly on-trend bull market moves), it puts on you an incredibly heavy burden of being right on timing. So the question is whether your out-of-sample confidence in your timing is so strong as to make you willing to pick the vehicle that is most demanding that you get it right, the one that gives you the possibly smallest margin for timing error. For quants who may want to try to create a laboratory downturn, it may be interesting to see how TLT stacks up against SH; in other words, can the -1.00 correlation do less damage to you as you wait for a timing model to catch up than a long position in TLT that goes sour due to bad timing. Another comparison is being off timing on TLT vs. SHY.
When these databases were developed, nobody cared. My guess is that the situation will be very different in 5-10 years and we’ll have easy access to convexity and duration. (This is the same problem we have with dividend details. Nobody carted when the databases were being designed and built, and now that people care, we’re finding ourselves burdened with trying to shove a round peg into a square hole).
Great post! And Great post again!!! Your last two post were great. Sincerely. And as I said I won’t be investing in TLT for the reasons you mention. Well said. I learned from all of your posts.
But if I change my mind, I might have to go to one of those radical online trading sites to include a long-term bond ETF in a packaged portfolio? You know like Vanguard? That is a funny!!! You are joking right?
Marc - while TLT may not perform as well going forward as the past, I urge that it NOT be removed from the hedge module. We need more hedge options, not less. With all the Fed Reserve talk about raising interest rates, the practicality of it is quite questionable as doing so adds to the US deficit. If interest rates were to return to “normal”, it would add something like $1/2 Trillion more debt per year that US citizens are on the hook for. And I believe recent studies from the Fed indicate that inflation won’t return to normal until 2020.
As for TLT needing to be part of a spot on market timing system, well I have to disagree with that. So long as TLT is part of a diversified set of strategies, including other market timing instruments, it can play a valuable role. And it isn’t a question of jumping into TLT when the stock market is going down the toilet, it can be choosing the lesser of two evils as I do in my Stitts Wealth Management systems. Timing is far from perfect but the results are still effective. TLT, along with the US Dollar, gold, all exhibit seasonal effects that can be taken advantage of. This is in contrast to SH or short SPY which does require spot on timing.
Please allow designers to continue using TLT and give subscribers the choice of whether or not to subscribe. R2G should be a free choice market, not a dictatorship.
As for market timing in general, one area to consider is the global carry trade. This is where the huge leverage ultimately resides and what dictates the so called risk on / risk off. A better understanding of the carry trade mechanism by which this occurs may produce some interesting timing systems. I have not done so, I’m just thinking out loud.
I am not an expert in bonds. I know the basics. But in my mind, we live in global economy. Germany recently issuing bonds with negative yields as of February 2015 (See this link ). US gov bonds are (for now) zero bound. But the relationship with zero is asymptotic. A bond that pays a yield (any yield) cannot under any circumstance go to a zero yield, regardless of how expensive the bond is trading.
Lets stop for a second to consider a negative coupon bond. That means that not only does the bond not pay a yield, it looses principal each month. I can’t even wrap my head around that. Who on earth would agree to such an arrangement? In the case of the German Bund… why not just hold Euro’s. That’s unless you’re betting the Euro won’t exist in 5 years when the bond’s reach maturity. But there are better ways to hedge that bet: Hold dollars!
Speaking of dollars. If I’m a German citizen looking for negative correlation, my money’s going to US Bonds. As the dollar strengthens - and it will strengthen even more IF and when the Fed raises rates - I get a second benefit to holding US bonds because the dollar is strengthening and the euro is weakening. All the more reason to think: global economy. With this scenario, you (any non-US investor) get a de facto carry-trade benefit when trading denominations back into your home currency.
So this all boils down to a question: What would cause a rational investor, or for that matter an irrational investor, to invest in German Bunds that is guaranteed to be worth less at maturity than they day it was minted? Euro-zone deflation? Nope. I’ll still hold currency. To my eyes, even the risky TLT sounds like a better deal than negative coupon bunds. Having said that, I wouldn’t put my greenbacks into TLT right now either. But I do think IEF would be a safe way to go for fixed income. For my personal portfolio, I am in IEF for recession protection. Rate rises will hit interim term bonds US bonds in the short run, but not nearly as hard as long term US bonds. The market will get to decide. Yield-curve surfing is probably in order for the next few years; bonds are still negatively correlated to equities. It might not be that way forever, but for the foreseeable future it looks like a very safe place to have my recession shock absorber $'s.
Just thinking out loud as a small retail investor. So rather than take a negative yield, I would go to the bank and and take a few (too few) thousand 100 dollar bills and hope that they would fit into a safe deposit box. Of course there would be a little loss on the safe deposit box.
Then again, in the US it would not be legal to take out all that money in cash–or would it just be paperwork and be on the FBI’s watch list for being a drug smuggler or something. Can you do that with a SEP-IRA without paying taxes?
In any case, it makes no sense to take a negative yield until you are forced to do so by the banks and government: which apparently is not too far fetched. Maybe if that happened in the US there would be zero (not negative) interest FDIC insured investments. Maybe: if so, up to how much?
Chris, yes, it is true. There are negative yielding bonds in Germany! Personally I would never ever invest in them but here is the rationale: When safety and avoiding drawdowns is becoming the number one concern, then you invest in the “safest” of all assets. And apparently some maket participants come to the conclusion, that this is German bonds. Germany is the strongest and biggest economy in Europe. Not because it is so good but rather because the other ones are a lot worse. Among the blind the one eyed man is king.
While the German government could of course also go bankrupt, it is a rather far fetched possibility under all reasonable assumptions.
Buyers of these negative bonds are usually not private investors but rather trusts, family offices and other large players. They invest a portion of the money to avoid big fluctuations and drawdown in assets in times of big turmoil (like when the Greek crises were most explosive). They don’t dare to invest in short-ETFs or other short insturments for reasons of risk aversion and lack of predictability. Even cash is eschewed because the Euro could sink in value. Under these assumptions there is nothing much left to invest in.
They figure: It is better to lose a little bit of money and be on the safe side than making some money but with high uncertainty and risk.
I also want to agree with Steve on TLT. No reason to remove it. There’s no risk of TLT going to zero due to a 25-basis-point increase by the FED.
If anything It would be good to allow designers to include more than one hedging instrument in a portfolio at equal weight. I’m working on a tax-smart R2G based on my existing R2G: ‘The Fundamentals - Massive Liquidity/Low Turnover’ that only trades on a quarterly basis. I’m considering having it always hedged with IEF. It would be nice to have it diversified with both a bond and gold component. I’ve tested this with a book with good results, but the book doesn’t really capture what I’m trying to achieve and can’t, of coarse, be offered as an R2G solution.
Negative yields are security/currency play. If you think USD is going to outperform say Yen or Euros over the time frame of investment, it make sense to buy a “risk-free” asset in the other currency. This has resulted in the aberrations observed in countries with negative rates. Negative rates in some countries beyond a few months are a play on the option value that if the Euro were to fail, you would be delivered your principal in say new DM or Guilders.
The demand for very short-term assets provided by financial institutions that do not have the ability to earn a spread net of their operational costs has pushed at least one Canadian bank, the Royal Bank of Canada, to offer its shortest term (1-29 day), redeemable commercial GIC at negative rates for their shortest term recently. I see the current rate at 0.001% but have a screen capture of the negative rate from a few months ago.
I think Jim Reid, DB’s credit analyst sums it up: “Our thesis over the last few years has basically been that the global financial system/economic fundamentals are so bad that its good for financial assets given it forces central banks into extraordinary stimulus and for them to continue to buy assets in never before seen volumes. The system failed in 2008/09 and rather than allow a proper creative destruction cleansing, policy makers have been aggressively propping it up ever since. This has surely led to a large level of inefficiency in the system which helps explain weak post crisis growth and thus forces them to do even more thus supporting asset prices if not the global economy.” (From the dire zero hedge.com today)
[quote]
Because TLT is inherently volatile, and because any gains it makes going forward will be counter-trend bear-market rallies (unlike in the past, when the gains were mainly on-trend bull market moves), it puts on you an incredibly heavy burden of being right on timing. So the question is whether your out-of-sample confidence in your timing is so strong as to make you willing to pick the vehicle that is most demanding that you get it right, the one that gives you the possibly smallest margin for timing error. For quants who may want to try to create a laboratory downturn, it may be interesting to see how TLT stacks up against SH; in other words, can the -1.00 correlation do less damage to you as you wait for a timing model to catch up than a long position in TLT that goes sour due to bad timing.
[/quote]The weekly volatility of TLT is roughly equal to that of SPY. See this screen.
Marc,
In your Forbes article you estimated a 50% loss for TLT over 20 years (if interest rates rise by 192% over 20 years from 2.6% to 7.6%.) That equals an annualized loss of 3.4% a year. Even in this drastic scenario a stock/TLT portfolio should make more than a stock/SH portfolio. That’s because SH would hopefully lose more than 3.4% a year over the next 20 years.
However, as you wrote in the article, you assumed that dividends were not reinvested. Reinvesting dividends at progressively higher and higher rates makes a huge difference. Looking at the historical record from this spreadsheet (which has interest rate data going back to the 1910’s), the two steepest twenty year rise in interest rates were from 1950 to 1970 and from 1961 to 1981.
Interest rates rose by 248% over the 20 year period ending May 1970–from 2.1% to 7.4%. Had you invested in TLT you would have been up (just barely) over this period.
Similarly, over the twenty year period ending Sep 1981 rates rose by 268% from 4% to 14.8%. TLT would have been up 1.9% a year over this period before fees.
Summary:
Fixed allocation to SH: -1 correlation, equal volatility to the market and a 3-6% annual cost.
Fixed allocation to long TLT: About -0.6 correlation, equal volatility to the market. (This means that TLT doesn’t always go up when stocks go down but when the hedge works it can overcome stock market losses. When both stocks and bonds fall at the same time–which is rare for good reasons–the portfolio overall can have a drawdown about as much as a 100% long portfolio. Furthermore, the TLT portion of the portfolio is likely to have drawdowns of about 25% during some bull markets.) TLT may cost somewhere in the range of -3% a year in a worst case scenario where rates climb very fast, but may make up to 2.5% profit a year if rates drift up slowly. (The drawdown–no pun intended–to using TLT as a hedge is not the overall returns and volatility of the portfolio but the loss aversion caused by TLT drawdowns.)
Fixed allocation to SHY: 0% correlation, almost no downside volatility and 0.5% annual profit. (While it is a steady source of income and can reduce volatility it loses money relative to inflation and it will not pick up the slack when stocks fall.)
I am ignoring the options of buy and hold treasuries directly to keep things simple.
I agree that even if TLT gradually loses money over long periods of time, it still can be a legitamate hedge. The purpose of the hedge is to smooth out returns of what you are hedging, in this case stocks. You want something that has low correlation, or is anti-correlated with your stocks.
For example, oil companies hedge against the price of oil. If 9 times out of 10, the hedge lost money, that is still okay. The 1 out of 10, it does its job.
Some hedge fund managers always put 1% of the portfolio in SP500 put options. Most of the time, this 1% is lost. But when the market crashes, this hedge kicks in.
SH was brought up as a hedge. If you have 50% SH, and 50% stocks, you have a market neutral portfolio. Great. However, SH is the inverse of the daily SP500 performance. It is not the inverse of the monthly, or yearly performance of SP500. Over long periods of time, SH is almost gauranteed to lose money, even if the SP500 is flat. This is true of all the inverse ETF, and even more so of the leveraged ones. Yet, I still think it is a legitamate hedge.
Not a new idea. As I mentioned in an earlier post in this or another thread, back in the early or mod 2000s, a Fed researcher communicated with me about this and shared an internal research paper. The issue, though, is political. If they tried to implement it through the economy as a whole, I suspect the Fed will cease to exist as an independent monetary authority, and that the fed knows it. If it could be done at all, it would have to be confined to rates on a narrow segment of savers/creditors – which might not be a half bad idea; it seems that judicial use of negative interest rates to unlock idle cash without penalizing savings that are needed might succeed. On the pother hand, holders of so-called idle or excess cash are most likely the major political campaign contributors – oh well.)
In any case, one who would use TLT in anticipation of such a thing would seem to have some strange notions on the nature of hedging and risk.
Interesting to read through this thread from 2015. @mgerstein was correct and TLT got slaughtered in 2000 to 2003. It was particularly bad for 2022 when stocks declined as well.
This is because the proponents of the 60/40 allocation sell bonds as "safe-havens". In reality, however, because Treasury returns are positively correlated with equity returns in both cross-country and long-term data, Treasuries are really just "low risk" rather than risk aversion.
His prediction was correct, but not for the right reasons. Widespread and substantial negative interest rates are not technically difficult, and the political impediments they face are potentially breakable in the face of a shock like the Great Depression. The real problem is that the return correlation between Treasuries and stocks is mostly positive.
