Investing in Bonds Versus Stocks
Are Long Term Bonds a Good Investment Now?
In this article I show data which I believe indicates that long-term bonds are an exceedingly poor investment that should be avoided at this time.
This is in spite of the fact that long-term bonds were a very good investment choice from 1980 and remained so until quite recently. But they are no longer a good investment choice.
The analysis here looks at the option of investing in (or continuing to hold) long-term government bonds rather than stocks for long-term investments at this time. For long-term we use 20 years.
The analysis here is based on U.S. data for stocks (S&P 500 index) and bonds (20-year U.S. government treasury bonds) from 1926 through 2014. The data source is a well-known reference book called “Stocks, Bonds, Bills and Inflation” 2013 edition — which provdes data through 2012. The book is published annually by Morningstar. (Ibbotson SBBI classic yearbook). Approximate data is used for 2013 and 2014.
A truthful but most dangerous observation
One of the most dangerous observations currently being made is that long-term bonds have performed well compared to stocks over the the past one year, five years, the past 10 years, the past 20 years and even the past 30 years.
It’s a true observation. But it is a highly dangerous observation to the extent that it implies that long-term bonds are likely to be a good investment compared to stocks going forward.
In fact, long-term bonds purchased today are almost guaranteed to under-perform stocks over the next 20 years.
It is important to understand that the return on a long-term government bond that will be held to maturity is known at the purchase date.
To analyze historic bond returns or expected future bond returns it is best to start with the simplest type of bond which is a bond that only pays off at maturity. These are called zero-coupon bonds because there are no annual interest payments (and because bond interest is sometimes referred to as coupon payments since bonds used to have interest coupons attached). Zero-coupon twenty-year bonds, for example, represent a lump sum to be received in 20 years. The lack of annual interest payments simplifies the analysis. Long-term zero-coupon bonds are purchased at a large discount to their face value and the interest is effectively received all at once when the bond eventually matures at its much higher face value.
In the past few decades, long-term zero-coupon bonds have provided excellent returns. A zero-coupon twenty-year U.S. government bond purchased in 1982 and held to maturity in 2002 ultimately returned precisely its initial yield which was about 14%. A zero-coupon twenty-year U.S. government bond purchased in December 1994 and held to maturity in December 2014 returned precisely its initial yield which was about 8.0%.
Regular bonds, in contrast, pay annual annual or semi-annual interest. A regular bond issued in 1982 and paying 14% annually ended up returning something less than 14% over its life since the annual interest payments (of $140 per $1000 bond) would have been reinvested each year at prevailing interest rates that turned out to be below 14%.
A zero-coupon twenty-year U.S. government bond purchased today and held to its maturity in January 2035 will return precisely its initial yield which is currently about 2.2%. A regular 20-year bond paying annual interest will also return something very close to its initial 2.2% yield over its life. This is because the 2.2% annual interest payments are so small ($22 per year on a $1000 bond) that it won’t matter much what interest rate those annual interest payments are reinvested at.
It seems obvious that 2.2% is not a good long-term return. And there are also very good reasons to think that stocks, as an alternative to long-term bonds, will return quite a bit more than 2.2% annually over the next twenty years. More about that below.
Historic returns of long-term bonds versus stocks
The following graph shows the compounded annual returns from investing one-time lump sums in 20-year U.S. government bonds (and rolling that investment over into new 20-year bonds each year, to maintain a constant 20-year bond maturity) versus investing the same lump sum in the S&P 500 index. The annual compounded returns for purchasing at each historic date and holding until the end of 2014 are shown.
To interpret this graph, let’s start with the right-hand end at 2010 and then move left. The blue line shows that a lump-sum investment in stocks made at the start of 2010 returned a compounded annual amount of about 15% per year through the end of 2014, while the red line shows that a lump-sum investment in 20-year U.S. government bonds made at the start of 2010 (and rolled over into new 20-year bonds annually, to maintain a constant 20-year maturity) returned a compounded amount of about 10% annually as at the end of 2014. We don’t show the results from investments more recent than the start of 2010 because returns over a short period of time are highly volatile and we are focusing here on longer terms.
Moving left we can see that these long-term government bonds have turned out, as at the end of 2014, to provide returns that were similar to and sometimes higher than investments in the S&P 500 for lump sum (as opposed to annual investments) money invested at the start of each year from 1994 to 2008. (The red bonds line is similar to and sometimes higher than the blue stocks line.) This would not have been anticipated by investors in those years, since stocks are generally expected to provide higher returns than bonds in the long term. And prior to that an investment in these 20-year bonds in the years 1980 to 1993 and held and rolled over annually into new 20-year bonds through the end of 2014, provided a return only moderately lower than that from stocks. Again this is not an outcome that would have been generally expected.
During the 1970’s however, history now reveals that buying stocks and holding through to the end of 2014 was (as would be expected) a better investment than going with and staying with 20-year bonds, although not by a huge amount.
Permanent investments in stocks prior to 1970 have turned out to provide a higher return, as at the end of 2014, than a permanent rolling investment in 20-year bonds and by quite a large margin.
Returns from investing in 20-year zero-coupon bonds and holding until maturity
The following graph shows the results for a one-time investment in a 20-year bond held to maturity compared an investment in stocks held for the same 20 years.
To interpret this graph, we can start from the left side. A one-time investment in a 20-year zero-coupon U.S. government bond in 1926 and held to maturity provided a compounded return of about 4% over its life. This 4% was known with certainty at the outset because that was the yield on the bond at issue. An investment in the S&P 500 made in 1926 and held for the same 20 years provided a compounded annual return of about 7%. But an important difference was that the 7% return from stocks was not known until the end of the 20-year period. An investment in stocks at the start of 1929 and held for 20 years ended up providing an annual compounded return of about 3.5% which was slightly lower than the return from buying a 20-year zero-coupon bond in 1929.
Investments in stocks in the years from 1932 to about 1954 and held for 20 years ended up providing returns that were far superior to buying a 20-year zero-coupon bond and holding until maturity. From 1955 to about 1980, stocks also provided a higher return over the next 20 years than did a 20-year zero-coupon bond held to maturity though by a smaller margin.
In the years from 1982 to 1992, investing in a 20-year zero-coupon bond and holding until maturity provided a return that was very similar to and occasionally higher than the return from stocks over the same 20 years.
The return from investing in a 20-year zero-coupon bond in 2004 and holding to maturity is shown at 5.0%. As of 2014, that bond will have provided a capital gain. But it will mature at par in the year 2024 and it will have provided a compounded annual return of precisely its initial yield of 5.0%. The return from investing in stocks in 2004 and holding for 20 years is not shown because it is not yet known.
I certainly suspect that when the returns from holding stocks for 20 years starting in all the years from 1995 through to 2014 are ultimately known, we will see that the blue stock line will almost always be above the red bond line.
The data for the returns from investing in regular 20-year bonds as opposed to zero-coupon bonds is not readily available but would be relatively close to the red zero-coupon line.
Why did long-term government bonds provide unexpectedly good returns in the past three decades or so?
Long-term government bonds can provide good returns for two possible reasons. But one of the reasons is only temporary.
The first and most important reason why a long term bond may provide a good return is that the initial interest rate paid by the bond turns out to be an attractive rate over the life of the bond.
20-year U.S. bonds issued in 1982 at 14% provided an excellent return (of precisely 14% annually for zero-coupon bonds and about 11% for regular bonds due to the reinvestment of annual interest payments at lower interest rates) if held through to their maturity in 2002 solely because that 14% was, in retrospect, a good return. Had we had hyper-inflation (as some feared at the time) then 14% might have been a poor return. But the 1982 (zero-coupon) bond provided a 14% return simply because that was what it paid. And we now know, in retrospect, that this was a good return over the 20 years from 1982 to 2002.
The second but temporary reason that bonds can provide a good return also came (temporarily) into play for the 1982 bond.
In 1983 the market interest rate on 20-year government bonds dropped to about 11% (from 14% in 1982). This provided a significant but temporary boost in the market value of the 1982 bond. The 1982 bond would have traded at a premium over much of its life as long-term interest rates declined significantly over the years. But in 2002 the bond matured at exactly its par value. The capital gain on the value of the 1982 bond was temporary and eventually the bond value declined to precisely its initial par value.
Over its full life the 14% return on the 14% 1982 zero coupon government bond was entirely driven by its contractual 14% interest rate. The decline in interest rates initially boosted its value but that was only a temporary impact. The fact that interest rates on 20-year bonds in 2002 had declined to 5.9% had no impact at all on the ultimate return at maturity in 2002 provided by the 1982 bond.
Bond Temporary Capital Appreciation
The temporary nature of market value gains on long-term government bonds is illustrated in the next graph which shows an index of the capital appreciation value of 20-year government bonds since 1926.
The blue line, plotted on the left scale, shows an index of capital appreciation on 20-year U.S. government bonds starting at 1.00 at the start of 1926. The index had risen slightly above 1.0 by the end of 1926, the first point on the graph. The index then rose significantly to 1.40 in the 1940’s. This meant that an investment permanently maintained in 20 year government bonds through annual rollovers to new 20 year bonds made in 1926 had appreciated in capital value by 40%. This excludes the value of the annual interest payments. This 40% increase was driven by long-term interest rates (shown on the red line plotted on the right scale) dropping from 4% at the start of 1926 down to 2.0% in the 1940’s. But this capital appreciation value gain eventually evaporated as the index returned to 1.0 when interest rates returned to 4% around 1959. And the bond capital value index slumped to about 0.50 in 1982 as long-term interest rates rose to 14%. This meant that the capital portion (which excludes the interest payments) of an investment in long-term government bonds made years earlier was worth only about 50% of the initial invested amount! The index then rose steadily all the way back to (not coincidently) about 1.40 as interest rates recently declined all the way back close to about the the 2% level of the 1940’s.
The red line, plotted on the right scale also shows precisely the return that would have been made by those investing in and holding to maturity a 20-year zero coupon bond in each year from 1926 to 2014. An investor in 1932 would have made 4%, in the 1940’s barely over 2%, at the peak in the early 80’s 14% and today’s investor in a 20-year zero coupon U.S. government bond held to maturity will, of a certainty, make 2.2%. Regular bonds are not zero coupon and therefore investors in regular bonds would have experienced somewhat different returns by reinvesting the annual interest payments.
Consider the long-bond issued at the end of 2010 and its misleading recent return.
An investment in 20-year government bonds at the end of 2010 returned a remarkable 28% in 2011. This was due to an equally remarkable decline in the market interest rate on these bonds from 4.14% to 2.48% during 2011.
Ultimately however, an end of 2010 (zero-coupon) government bond is going to return precisely 4.14% compounded per year over its 20 year life. The capital gain due to an interest rate decline in 2011 provides only a temporary gain that will be reversed. The 28% gain is largely irrelevant to an investor that holds the 2010 bond to maturity in 2030. It is only relevant to bond traders that have sold or will sell the bond prior to the capital gain reversing.
What Return can we now expect from 20-year bonds?
A 20-year U.S. zero-coupon government bond purchased today should be expected, over its full life, to return its current yield of 2.2% per year. If 20-year interest rates soon decline the bond will provide a temporary gain in market value. If interest rates increase it will suffer a temporary loss in market value. But over its life this bond will return only and precisely 2.2%.
The fact that, as shown in the first graph above, an investment in 20-year bonds made, and permanently maintained in 20 year bonds, at any time from about 1968 to 2008 has to date returned a compounded return of about 8% or more is completely irrelevant to an estimate of returns going forward. Some of that return, will prove to have been temporary as bonds now valued at well above par eventually mature at only par value. Far from recurring in future, this temporary return boost will reverse in future years. Much of the 8% or more return has occurred simply because bond interest rates, over the past few decades, were much higher than today.
It would be a huge mistake to assume that a twenty-year 2.2% bond issued today will ultimately earn (over its full life) anything close to the approximately 8% returns bonds achieved to date for investors who invested in all the years from 1968 to 2008. If you base your bond return expectation on the high average bond returns made (to date) by investing at any time in the past 47 years you will implicitly be making a seemingly logical but actually completely flawed assumption.
Technically, the return from a 2014 20-year regular government bond will be a little bit different than precisely the 2.2% initial yield if interest rates change. If interest rates rise there will be an opportunity to reinvest the annual interest payments at a higher rate. Or, if interest rates decline the reinvestment will be at lower rates which would lower the 2.2% return. However with the annual interest coupons on a $1000 bond being a meager $22, the impact of reinvested interest is minor at today’s low interest rates.
Should we invest in long-term government bonds?
In my view the data indicates that the answer is “NO!”. Not unless you are satisfied with an expected return on the order of 2.2% for 20-year government bonds. And long-term higher rated corporate bonds also will return no more than about 3.7% if held to maturity, since that is their approximate current yield.
Why should we expect Stocks to Return more than bonds?
The wrong way to predict stock returns would be to look at the return since year 2000 of (a highly volatile) average of about 4% per year or to look at the long term historical return of about 10% per year.
Mathematically the return from stocks will equal the dividend yield plus the rate of growth in earnings. (This assumes the P/E ratio will remain constant.) The dividend yield on the S&P 500 is currently about 2.0%. If earnings per share grow at about the rate of GDP, say 2% real plus 2% for inflation, this would suggest that stocks will return about 6%. Although this is low, it easily beats the current known return on 20-year government bonds of 2.2% and the known 3.7% or so return from 20-year high-grade corporate bonds.
In 20 years 2.2% turns $1000 into $1,545, while 3.7% turns it into $2,068 and 6% turns it into $3,207.
Implications for Investors, Including Pension Funds
Twenty-year U.S. government bonds purchased or held as of early 2015 are destined to return only about 2.2% over their lives. Stocks, meanwhile are providing dividend yields of 2.0% on average and the dividends and earnings can reasonably be expected to grow at 4% or more for a total return of 6% or more over the long term.
I believe that history will show that pension funds and other investors that make large allocations to (or even continue to hold large allocations of) long-term bonds in early 2015 are making a serious mistake. They would be better off to hold cash and short term investments rather than long-term bonds.
Pension funds and other large institutional investors are blindly following their historic asset allocation percentages and are ignoring common sense. Long-term bond returns have been very good in the past three decades because long-term interest rates started out high and have dropped. But that same drop guarantees that bond returns, from an early 2015 investment in 20-year bonds will be very low indeed over the next 20 years. And one has to be very pessimistic to forecast that stock returns will fail to materially exceed these low bond returns.
Money invested in a well diversified portfolio of stocks in early 2015 is almost (but never quite) certain to exceed the return from investing in long-term government bonds in early 2015 which will, of a certainty, be in the range of 2.2% if held for the next 20 years.
In the short term, bonds may do better than stocks. But stocks will almost definitely do better than bonds over the next twenty years.
The suggestion to avoid long-term bonds at this time violates the traditional advice to always maintain some exposure to long-term bonds in your asset allocation. My belief is that following a traditional asset allocation approach at a time when interest rates are near the lowest levels in history defies common sense. History will be the judge.
Again, note that this article says nothing about holding cash or short-term bonds, it only compares long-term government bonds with stocks.
Shawn Allen, President
January 18, 2015
The original version of this article was written in 2012 when 20-year U.S. bond interest rates were 2.4%. As of today following the advice of that article to avoid long-term bonds would have been good at the time it was written but perhaps not so good if followed at the start of 2014 when interest rates had increased to 3.68%.
We can turn to Warren Buffett for some support for our arguments above.
In his 1984 letter, Warren Buffett wrote about the irrationality of investors buying long-term bonds at times of very low interest rates.
“Our approach to bond investment – treating it as an unusual sort of “business” with special advantages and disadvantages – may strike you as a bit quirky. However, we believe that many staggering errors by investors could have been avoided if they had viewed bond investment with a businessman’s perspective. For example, in 1946, 20-year AAA tax-exempt bonds traded at slightly below a 1% yield. In effect, the buyer of those bonds at that time bought a “business” that earned about 1% on “book value” (and that, moreover, could never earn a dime more than 1% on book), and paid 100 cents on the dollar for that abominable business.”
“If an investor had been business-minded enough to think in those terms – and that was the precise reality of the bargain struck – he would have laughed at the proposition and walked away. For, at the same time, businesses with excellent future prospects could have been bought at, or close to, book value while earning 10%, 12%, or 15% after tax on book. Probably no business in America changed hands in 1946 at book value that the buyer believed lacked the ability to earn more than 1% on book. But investors with bond-buying habits eagerly made economic commitments throughout the year on just that basis. Similar, although less extreme, conditions prevailed for the next two decades as bond investors happily signed up for twenty or thirty years on terms outrageously inadequate by business standards.”
Buffett described buying tax-exempt bonds at yields around 1% in 1946 as in effect the purchase of an abominable business. And he said the bond investors accepted terms that were outrageously inadequate by business standards for the two decades after 1946. I don’t know what tax exempt bonds paid during that period but 20-year government bonds yielded 2.4% to 2.0% during 1946. Today, the 20 year yield at 2.2%, is precisely at the level where Buffett considered an investment in long-term bonds to be similar to the purchase of an abominable business offering terms that are outrageously inadequate by business standards. And today, the S&P 500 trades at a trailing earnings yield of about 5.3% (the inverse of its P/E ratio of 19.0), which is vastly higher than the bond cash yield. It is true that the stock earnings yield is not available in cash (although about 2.0% of it is as dividends). The remaining 3.3% earnings yield is retained by the companies for reinvestment, often at double digit ROEs, for the future benefit of the share owners.
It would be a mistake to invest in long-term bonds today on the basis that they have provided returns similar to stocks for many years. We know, of a certainty, that 20-year government bonds purchased today, and held to maturity, will provide meager returns around 2.2%. And we can rationally expect stock returns, based on the S&P 500 index (with 2% dividends and another 3.3% retained for reinvestment, at relatively high ROEs) to be higher, over the next 20 years, than these inadequate bond returns.