Tuesday, September 27, 2016

Shrinking Bonds

As I write this, Canadian 30-year bonds yield 1.674% per year. If we assume that Canada will be able to maintain its 2% inflation target, these bonds will lose purchasing power for the next 30 years. Investors today are willing to tie up money for 30 years and get back less at the end of it all. This seems remarkable to me.

Of course, investors can choose to sell these bonds before they mature, but if we follow the life of an individual bond, its owners will share the loss of purchasing power for 30 years. Some investors may hope yields will drop further creating a capital gain. Some may think inflation will drop or that we might even have deflation. Others may think that alternative investments, such as stocks or real estate, will fare worse.

I don’t know if bond investors are being rational, but I find it hard to look past the apparent near certain loss of value. Over 30 years I’m hoping my stocks will roughly triple in real value and that my house will hold its value. I may be disappointed, but it’s hard to buy bonds with disappointment seeming to be a near certainty.

Bonds have the virtue of reducing portfolio volatility. This is useful for investors who can’t handle too much volatility or who have enough money that they don’t need to take much risk. However, these benefits aren’t enough to get me past the thought that buying bonds today is a money-losing proposition.

33 comments:

  1. "... Over 30 years I’m hoping my stocks will roughly triple in real value and that my house will hold its value. ..."

    Totally agree with the house holding value. Renovations offset depreciation. Most capital appreciation in housing come from land appreciation -- the dwelling rarely appreciates except in housing bubbles.

    House value (real) = capital (start year) + inflation - depreciation ~= capital (start year)

    ReplyDelete
    Replies
    1. @Anonymous: I'm not expecting the combination of my house and land to outpace inflation, but who knows what will happen. It seems much more certain that bonds will lose ground to inflation.

      Delete
  2. Michael. Do you plan to hold mostly equities in your nest egg at the start of your retirement or withdrawal phase? If so, how do you plan to deal with "sequence of returns" risk?

    ReplyDelete
    Replies
    1. @Garth: My current plan for retirement is to have 5 years worth of spending in cash or GICs and the rest in stocks. If the stocks perform poorly, then I spend the cash a little slower to give the stocks some time to bounce back.

      Delete
    2. The cash bucket will need to be refilled at some point, so it then becomes a market timing issue as to when to do so. Long bear markets could be hard to work around...

      Delete
    3. @Garth: I don't plan to leave any discretion. The refilling will take place annually. If stocks are 5% below target, then my spending drops 5% and the 5-year cash cushion only needs a top up of 75% of the amount the previous year.

      Delete
    4. 30 year bonds seem unattractive but who knows? I'd be more inclined to use 5 year bond ladders.

      I was inclined to a similar retirement income strategy (5 years spending, the rest in stocks) until I read http://livingoffyourmoney.com/. Its similar to the "bucket" strategies which doesn't perform very well by a number of measures. The general conclusion of the book is you can get a lot more and even retirement income with income harvesting strategies that have an initial bond allocation of more like 50%. Putting less that 40% in bonds at the start gives you a high probability of frequently having lower income than necessary or running out of money. Not having enough in fixed income puts you at risk of having to sell stocks when they are down, and that is what kills a retirement income plan and income plan fundamentally different than savings plan.

      Delete
    5. @Greg: What worries me about bucket strategies is that most seem to rely on some amount of market timing. I want a purely mechanical strategy.

      I've generally been disappointed by how little good information there is about decumulation. I'll give the book a read.

      Delete
    6. @Greg: I'm about 100 pages in and am still mulling, but I have one serious concern about the fixed withdrawal strategies. For comparisons based on "MSWR-100%" (meaning a maximum safe withdrawal that works for all 30-year periods without ever running out of money), all but the worst strategies the book considers perform worst for the 30 years starting in 1969. This means that this one period determines which strategy is best and all other periods are irrelevant. The best strategies are then likely tuned to the particular pattern of stock and bond returns for this period.

      Delete
    7. Michael, chapter 3 is actually about "income harvesting" strategies, not withdrawal strategies. Comparing the rate the income is harvested with MSWR and evaluating them based on a fixed withdrawal strategy is useful in evaluating them, but the withdrawal strategies are covered separately in chapter 4. The recommended withdrawal strategy is "Extended Mortaility Updating Failure Percentage" where you withdraw a percentage that is based on your actuarial life expectancy, with some extra rules to provide a floor and ceiling around the inflation adjusted initial withdrawal amount to smooth out income over time. The net effect is you save a little money when markets are good to cover for when markets are bad.

      Delete
    8. @Greg: Come on now. Chapter 3 is about fixed withdrawal strategies as I said. Calling them "income harvesting" changes nothing. If you're saying that nobody should actually use a fixed withdrawal strategy, then I agree.

      I'm hoping that the variable withdrawal strategies of chapter 4 don't suffer from the problem I identified with chapter 3.

      Delete
    9. Oh, and he also takes care to make sure the strategies aren't overly trained on US markets, he tests them on Japan, Europe, and simulated markets with bootstrapping techniques. There is a lot of evidence that the best strategies should be able to handle the worst times to retire like in the late 60's without running out of money and letting yearly income get too low. But the best strategies are efficient, they provide more income in better retirement periods, a "harvesting rate efficiency" metric is introduced to compare how efficiently withdrawal strategies capture the income that is available for a retirement period.

      And later on the "harvesting ratio" which is kind of like the sharp ratio for retirement income portfolios to evaluate which portfolios provide the best trade off between retirement income and risk. There is some convincing evidence that more that 60% stock isn't very safe, even for really long- 50 year- retirements.

      Delete
    10. @Greg: The author did all the things you describe quite well. However, there is still the problem of tuning the strategies to a single 30-year period. This doesn't prove the recommendations are bad; it just casts doubt on the evidence they are good.

      Delete
    11. Chapter 3 is about income harvesting strategies and comparing them based on how they would do with a fixed withdrawal strategy. Income harvesting definitely isn't another name for withdrawal strategies according to the book, take a look at the "Modern Mechanics" section on page 15.

      I'm not sure you've identified a problem with chapter 3 at all. Don't you think that all that simple-block bootstrapping simulation and testing against different markets from what the strategies were originally designed for gives a good basis for comparison and removes data mining bias toward 30 year retirements starting in 1969?

      Delete
    12. @Greg: I'm used to the "strategy" part of a "fixed withdrawal strategy" referring to the rules for choosing which assets to hold and which to sell. The author chooses to split this part out and call it an "income harvesting strategy".

      Using bootstrapping and other data sets helps somewhat but not enough. If I create a strategy that is inferior for all other starting years, but is slightly better across the data sets (and bootstrapping) for 1969, then my strategy would win. The next bad 30-year period is likely to have different world-wide stock return patterns than we had starting in 1969. The decision of which strategy is best is based on too little data. But this isn't proof that the recommended strategies are bad.

      Delete
    13. I would agree if a strategy was inferior for starting years other than 1969, but backtesting on the recommended strategies shows they are more often than not better across all retirement years and never much worse than the annual rebalancing baseline. The bucket strategies are uniformly worse than annual rebalancing, the leave a lot of accessible income on the table when market conditions are good for retirement (and leave bigger inheritances as a result). But if you are trying to optimize for maximum safe income, a large inheritance is suboptimal.

      Bootstrapping with 1000 simulated markets will have about 10 with conditions as bad as 1969 but different patterns, so the bootstrapping tests should show if a strategy is unduly tuned to tuned to 1969.

      There is a big Boggleheads forum discussion of this book https://www.bogleheads.org/forum/viewtopic.php?f=10&t=192105. I haven't seen much convincing criticism there, the biggest detractors seem to have a lot of vested interest in strategies they devised but don't have the rigorous testing to back up their claims.

      I guess you can tell I found the book quite compelling :).

      Delete
    14. @Greg: I'm trying to keep an open mind, because, as I said, I haven't finished the book. Giving superior results in other time periods certainly lends some credibility.

      I have a number of concerns about just about all strategies proposed:

      1. Historical returns support 30-year retirements with starting withdrawals in the four-point-something percent range. What if future returns are uniformly lower (as many experts predict)? This means that if I start with the best strategy, I'll probably be spending too much initially and am destined to have to live on less eventually. This is particularly troubling if I end up with a longer than 30-year retirement. There is little doubt that the withdrawal percentages suffer severely from data-mining.

      2. Wouldn't it be better to develop strategies taking into account actuarial statistics rather just picking some retirement duration? Maybe I'm the guy who will live to 115.

      3. If it doesn't make sense to own more than 60% stocks, shouldn't this rule carry back to before retirement? After all, a high percentage of us will be forced to quit work for health reasons or because our employers will fire us.

      Delete
    15. I think you'll find answers to your questions later in the book. Some previews: The recommended initial withdrawal rate is tilted based on current market valuation metrics such as CAPE. If markets are uniformly lower in the future, that is what the book calls "speculative risk"- things that haven't been seen before or invariants in historical data that stop in the future. The only way to reduce speculative risks is to be conservative and stress test strategies with market conditions worse than we have ever seen by back testing with scaled down market returns. The book does 40 year retirements with with a scaled down market (and wrapping 2010 followed by 1928 to get more 40 periods to test against) where the MSWR-100% is 3.1% The recommended strategies automatically adjust to work efficiently and have an average withdrawal rate of 4.8% (inflation adjusted of the initial withdrawal) and withdrawal can drop into the low 2% range for several years if your retirement coincides with a particularly bad portion of this scaled down market. Interestingly, the author uses tests on this scaled down market to decide not to recommend the income strategy that he invented and performed best on non-scaled down markets could ran out of money a few times on this scaled down market for 40 year retirements. The recommendation is to make sure 2% withdrawals cover your basic income needs don't necessarily spend every penny that the strategy allows you to withdraw when markets are good.

      The recommended income strategy does use actuarial statistics to set the withdrawal rate and there are tests for retirements up to 50 years long. Longer than 50 years is actually not much different than 50 years, just use 50 years left as the parameter if you want longer than 50 years. The book also has recommendations on how to use annuities to insure against longevity risks as an option.

      Regarding 3, I think you'd have to be pretty conservative to have 60% or less in stocks in the wealth accumulation phase of your life before retirement. When you are young health issues are less likely and you should have disability insurance, and it is unlikely that you lose your job and never find any job ever again. If you plan to have 40% in fixed income at retirement I think it's a good idea to be getting there as you approach retirement, rebalancing in the last 5 or 10 years you plan to work for example.

      Delete
    16. @Michael
      Here is another withdrawal strategy you should have a long look at.

      http://www.finiki.org/wiki/Variable_percentage_withdrawal

      Delete
    17. @Garth: Many withdrawal strategies can be expressed in a form that gives a portfolio percentage based on age and possibly other factors. I expressed my own strategy based on what I call cushioning in this form. (See the link to the spreadsheet in http://www.michaeljamesonmoney.com/2014/02/cushioned-retirement-investing.html)

      Delete
  3. Yes, this seemingly more permanent low rate reality is far reaching and raises issues for retail, pension, etc. bond investors.

    It might be interesting to understand who are the purchasers of these l/t fed govt bonds and why? Maybe corporate bonds will have to play a bigger role for investors with marginally better yields.

    Bonds seem to work in 30yr or so cycles and we appear to be in the earlier stages of a record low cycle. I wonder if global bonds could implode.

    My guess is people are likely to be disappointed with bond "real returns" for many years, and the same could be true with stocks. However bonds are much closer to a disappointingly low certainty than stocks.




    ReplyDelete
    Replies
    1. @RBull: I don't know who are the main owners of these long-term bonds. Certainly, individuals own many bonds through mutual funds and ETFs. But do they dominate?

      Delete
    2. Hmmm, I'd be careful about corporate bonds. I was looking into why there has been so much activity with companies buying back their shares. I saw some indication that companies are taking advantage of low interest rates to issue corporate bonds, and taking this borrowed money to buy back shares. It's all starting to look like another pyramid scheme that will end badly. Having said all that, stocks are still the only game in town to get some return in the long run.

      Delete
    3. Yes anyone buying corporate bonds needs to be careful, like all investments. Although I would suggest if you believe it is all a pyramid scheme you want to be careful with equities, especially since bonds are senior.

      Delete
    4. @mj
      Good question that I don't know. I wonder is this foreign investors that are now upside down with rates, and/or is it pension funds etc? I checked VAB for example that has a YTM of 1.7% and ~14% of distributions is from 25+year bonds.

      Delete
  4. I like your writing a lot and purely playing Devil's Advocate here:

    I think you would be among the first to argue that "you cannot see the future". For that reason you don't pick stocks, etc.

    Isn't proclaiming that bonds represent an "apparent or near certain" loss of value doing just that? Sure, it's hinged on some mathematics and seemingly safe assumptions - but isn't that the same thing stock pickers do?

    Who says we are not headed for deflation? Or that we wont ever see sustained 2% inflation over the next 30 years? Personally, based on history, I think these things unlikely - as do you - but I don't know. Thus for the same reason I cannot pick stocks, I cannot exclude bonds.

    ReplyDelete
    Replies
    1. @Anonymous: As I've said, I don't think it makes sense for everyone to abandon bonds. So, I think we're mostly in agreement. However, I don't think what I've done is similar to stock-picking. Stock pickers are trying to pick a mix of stocks that will outperform. But only a tiny fraction of stock subsets will outperform by enough to cover costs. So, they are trying to find a needle in a haystack. I observed that bonds losing real value is a high-probability (but not certain) event. That said, it makes sense to protect against a reasonable range of possible outcomes. The lowest 30 years of U.S. inflation since 1913 averaged 0.7% per year (1920 to 1950). So, we can reasonably say that 30-year bonds could have positive real returns. And we can also say it's possible (but unlikely) for stocks to underperform bonds for the next 30 years. I think my remarks are a good reason not to own 100% bonds for 30 years. But I didn't intend to argue that everyone should abandon bonds altogether.

      Delete
  5. Interesting interview with William Bernstein here
    http://www.aaii.com/journal/article/investing-to-avoid-the-consequences-of-being-wrong.mobile

    He talks a lot about how true risk changes with age...

    ReplyDelete
    Replies
    1. @Garth: Bernstein is always intelligent and entertaining. When I run out of human capital, I'll have to suck it up and shift assets to fixed income. But if yields are still so low by then, it'll make my stomach hurt.

      Delete
    2. Like Bernstein says, all you can do at that point is be philosophical. :)

      Delete
  6. I think a good way to think about bonds is that we don't own them for return (we own equities for that), we own own bonds to to reduce the likelihood of permanent loss of capital by reducing the volatility of the portfolio so we don't panic sell in a crash. For those who don't need that smoothing, there may be no need to own any bonds.

    As for the term of the bond, many authorities suggest not owning bonds with a term of greater than 10 years as at that point you get more risk than return. Certainly a total bond fund such as VAB does own some longer term bonds, but they own many more short term bonds, so the duration of the fund is close to an intermediate term bonds fund of around 6 years.

    ReplyDelete
    Replies
    1. @Grant: All that is true, but valuation does matter at some point. If we take it to the extreme of a -100% return, bonds would still provide the same volatility-reducing benefits. I don't think the current gap between expected bond and stock returns is large enough to take us into extreme territory, but it's something to watch for.

      Delete
  7. The exchange below is reproduced to remove broken links.

    ----- Phoebes September 27, 2016 at 8:34 PM

    Is your argument also relevant for bond ETFs? Thanks for your blog - it has helped me got out of the bank advisor/high MER trap.

    ----- Michael James September 28, 2016 at 7:52 AM

    @Phoebes: I'm glad you've found my thoughts useful. I was talking about 30-year government bonds. Bond ETFs usually hold a mix of durations and may also hold higher-yielding corporate bonds. For the bond ETFs sticking to government bonds, it's still true that their yields are below the inflation target. But that doesn't necessarily mean it's a bad idea to own some bonds. Stocks have a nasty habit of crashing at unpredictable times, and too many investors lose their stomachs for the roller-coaster ride and sell low. Bonds provide stability.

    ReplyDelete