Major Flaw #1: Interest Rates Have to Rise More than People Expect for Treasuries to Underperform Cash

Earlier in the book I emphasized the importance of breaking down returns into their core components. The return of cash plus the excess return above cash make up the total return of an asset class. Since the return of cash is risk-free, you should focus on the excess return above cash when analyzing asset class returns. You can get the cash rate with no risk, so if you are going to take risk you need to understand how much you can expect to be compensated.

When you decide to exchange cash for a long-term Treasury bond, you are effectively making a bet that you will be compensated a little more than what you would earn if you just held cash over time. Since both cash and Treasuries are considered risk-free in terms of principal value, you would only take the price risk associated with Treasuries if you were compensated with excess returns above cash. Since Treasuries are merely a longer-duration version of risk-free cash—because they, too, do not contain credit risk—the excess returns Treasuries offer are understandably low relative to other, riskier asset classes.

The actual excess return above cash that you will receive with your Treasury investment, like all other asset classes, depends on how events transpire relative to market expectations. The future discount is relatively straightforward to discern for Treasuries because it is reflected in the slope of the yield curve. Specifically, the yield curve tells us the market's expectation for the future path of cash rates. An upward sloping curve demonstrates the discounted increase in cash rates over time. Cash rates and short-term interest rates are nearly identical and are set by the Federal Reserve. The steeper the slope, the faster short-term interest rates are expected to rise.

Therefore, when you think about whether Treasuries are attractively priced relative to cash the key question to ask is whether you expect interest rates to rise faster than what is currently priced in. If you do, then you should underweight Treasuries (relative to an efficient, neutral mix); and if you don't, then you ought to hold more. For example, if cash is yielding 0 percent and long-term Treasuries are offering a 4 percent yield, then that implies that interest rates are expected to rise significantly over time. If they do rise, but less than priced in, then you would be better off buying Treasuries than holding cash. In this case Treasuries would earn excess returns above cash because cash rates rose less than expected.

From this perspective it should be apparent that contrary to the popular view, the current yield by itself does not matter. You need to look at the entire yield curve to identify what future conditions it is discounting. A 4 percent long-term Treasury yield means something very different when cash is yielding 0 percent versus when it is yielding 6 percent. At 0 percent, the yield curve is extremely steep and at 6 percent it is inverted (signifying the anticipation of falling short-term interest rates). In both cases, the excess return above cash is dependent on the future path of interest rates versus the discounted levels. In fact, since 1927 the correlation of future excess Treasury returns and the starting yield is just 10 percent or 0.1 (meaning there is effectively very little correlation between the two). The excess return depends on how the future plays out relative to what was already priced in (and you know that the odds of it going one way or another are about fifty-fifty).

To illustrate this point, let's review the yield and returns of Treasuries during the past six years. The 10-year Treasury fell below 4 percent in December 2007, at which time nearly every bond investor concluded that interest rates were near historic lows and destined to rise. Shortly thereafter at the onset of the financial crisis, short-term interest rates fell to almost 0 percent. With the framework described above, the appropriate way to have analyzed interest rates was to consider the steepness of the yield curve to gain a more complete perspective of what the low rates meant. Since everyone expected rates to rise, this market view was already reflected in the yield curve with low rates at the short end (the end closest to the present) and much higher relative rates at the long end (the end farther out).

A well-informed investor would have asked what has caused short-term rates to fall. Historically, interest rates have gone down to zero only in times of great distress and when there is need for massive monetary stimulation. The last time interest rates were near zero in the United States was during the Great Depression, and they stayed there for 15 years, from 1933 to 1948. In Japan, interest rates approached zero in 1995 and have remained there ever since (19 years and counting). Monetary policy dictates lowering short-term interest rates to stimulate growth. When interest rates are lowered to zero, that suggests that the economic environment is so fragile that zero rates are warranted. In today's environment, zero rates are in fact insufficient to stimulate growth because of the overly indebted balance sheets of many developed economies (the United States included). Thus the question to ask is how long short-term interest rates may stay near zero given the economic environment, as opposed to just automatically concluding that they must rise just because they are low. The timing is crucial because it ultimately drives long-duration bond excess returns to a greater degree than the starting yield.

The reason Treasuries performed so strongly was because everyone expected rates to rise rapidly, and instead rates fell. Why did they fall? Economic growth underperformed levels discounted in the market, which were far too optimistic given underlying economic conditions. This is a major oversight of the vast majority of investors.

In reality it is not the low yield of Treasuries that has been unusual but the near zero yield of cash, which rarely occurs. Since all asset classes price relative to cash, low yields impact everything. They are just more easily observable in low Treasury yields. This is a crucial insight: low cash yields lower the total returns of all asset classes since they are all priced relative to cash. The excess returns should be your focus, and they remain unchanged as a result of low-yielding cash. If you observe every asset class relative to cash as described in an earlier chapter, then this thought process should make sense. This point is most obvious when considering Treasuries because of the close similarity between this asset class and cash.

Another argument against holding Treasuries in a portfolio is that the yield is said to be artificially low because the Fed is printing money and buying these bonds. The thinking goes that this is not normal and is in effect artificially depressing rates far below where they would be were it not for the government's manipulation. And, once the Fed stops doing this, rates are surely going to rise. Thus, in today's unique environment, the argument continues, it is not just how the future transpires relative to expectations but the role government plays in the process—it is not a free market, but one that is being controlled by a noneconomic participant. Many experts have proposed this line of thinking.

However, the flaw in this perspective is that it fails to fully appreciate the reasons for the Fed's involvement. Think of our discussion in the first chapter. The deepest economic collapse since the Great Depression forced the Fed to first lower interest rates to zero and then engage in quantitative easing (or printing money) to push longer-term interest rates lower and inject money into the economy. The last time this was done to this degree was during the Great Depression, when similar economic conditions existed. In other words, interest rates are low because the economy is extremely depressed, which is the type of environment in which Treasuries are biased to do well (and they have).

In fact, it is debatable whether interest rates would be higher or lower if the Fed were to stop buying Treasuries. The first few times Treasury purchases ceased shortly after 2009, interest rates actually fell even lower because the lack of printing of currency caused the economy to weaken, which caused rates to fall further. The forces of the economic environment outweighed the reduced buying by the Fed to force rates even lower. Conversely, each time the Fed started to buy bonds, rates actually rose because the economy strengthened. Thus, the answer to the question of whether rates are artificially low is unclear. Had the Fed never bought a single Treasury bond interest rates might be significantly lower than current levels because of the probability of an economic collapse. Perhaps interest rates are artificially high! Consider, for instance, that several European countries have interest rates lower than those in the United States even though the ECB, as of this writing, has printed far less than the Fed.

Major Flaw #2: Long-Term Treasuries Can Deliver Sizable Gains over Shorter Periods

Please note that in this section when I refer to owning long-term Treasuries, I am suggesting maintaining a constant duration to these bonds rather than buying a bond and holding it to maturity. This is an important distinction. If you buy a 10-year Treasury bond, for example, and you hold it until it matures, then the return you will earn is known at the time of your purchase. It will simply be the yield to maturity of the bond. This is the return that most investors think about when they consider investing in Treasury bonds. This logic may sound intuitive because when you analyze the potential return of a bond it seems to make sense to look at the yield it offers plus a return of your principal. From this perspective, a low yielding Treasury bond would understandably look quite unappealing.

However, if instead you buy that same 10-year bond today and then after a year sell it and buy another 10-year bond and continue the process indefinitely, then you will maintain a constant long-maturity Treasury bond. The return stream of this portfolio may be different from that in the first example when the bond is held to maturity, depending on the economic environment.

When considering Treasuries, most investors are concerned that the potential upside is limited when the yield is low. Many are also concerned that there is significant risk of loss because rising interest rates means that Treasuries will lose money. These commonly held views are simply not true when considering the excess returns above cash. We already know the return of cash is low, but this reality lowers the returns of all asset classes, not just Treasuries. The focus should be on excess returns above cash, as I've reiterated multiple times.

The excess return of Treasuries, like other asset classes, is based on how the future transpires versus what was discounted. If interest rates are priced to rise rapidly and they actually fall, then significant returns can be expected. This is because of the huge mismatch between what happened and what was anticipated to occur. You can see that when nearly everyone expects a certain outcome, the pricing in of that eventuality actually creates an opportunity for significant outperformance because of the extreme view. On the other hand, if few expect rates to rise and they do rise because of the shifting economic conditions, then that would lead to underperformance in Treasuries. Treasury outperformance versus cash is possible, then, even with low starting yields.

There is one condition, however: To achieve high returns investors must own longer-duration bonds. Duration is key. It is duration that can produce the high returns and it is duration that helps build balance. Investors have been reducing duration because they are afraid of losing money. Instead they should be increasing duration, not because they expected interest rates to fall, but to improve the balance in the portfolio. If rates happened to rise more than expected, then other segments of the portfolio could have benefited (such as equities, if rates rose because the economy was strong). If rates fell (as they did), then the Treasuries would have appreciated enough to offset the losses in the other asset classes. This critical point will be fully explored later in this chapter and in the subsequent discussion about the makeup of the balanced portfolio at the end of the book. The key argument I am trying to make at this point of the discussion is that you can achieve very good excess returns even if the starting yield is low—if you maintain long duration.

Once again, recent experience provides a helpful example of the dynamic described above. The 30-year Treasury yield on June 30, 2011, was 4.4 percent and it closed at 2.9 percent just three months later. The reason it fell so much was because the economy materially weakened unexpectedly during that short time horizon because the Fed stopped buying Treasuries. Since the duration is so long, the excess return during that three-month period was 30.9 percent! The S&P 500 dropped 14.1 percent during that three-month span and investors who didn't own long Treasuries were fully exposed and likely lost significant capital.

What about the consensus view among experts that going forward it is highly unlikely that interest rates will fall again? The problem with this argument is the simple fact that the consensus view has been an extremely poor indicator of future returns time and again. Consider that the three best years in the history of the Treasury market all occurred at a time when Treasuries were most out of favor (+32 percent excess returns in 1982 after an 18-year run of rising interest rates, +39 percent in 2008, and +35 percent in 2011 when interest rates were widely considered to be too low). Not only did Treasuries outperform consensus expectations, but they emphatically delivered record returns!

What about the argument that bond yields do not sufficiently compensate investors against inflation? The current yield may be less than expected inflation, but that does not necessarily mean that the total return (yield plus principal gains) will be less than actual inflation. In terms of total return, the example above clearly demonstrates that with enough duration, the total return can far exceed inflation. Furthermore, we do not yet know what inflation is going to be over time. It may be much higher or it may be much lower (or even negative as it was for a moment in 2008). The key consideration is whether the future plays out as expected. Far too often, investors are guilty of placing too much confidence in their ability to accurately predict the future despite having a subpar track record (selective memory in this area tends to obscure reality).

Major Flaw #3: Cutting Treasury Exposure Reduces Balance and Makes Portfolios More Risky

Many investors have been cutting their exposure to Treasuries as a response to their concerns about limited upside and big downside risks in this asset class. The act of reducing Treasury exposure can take two main forms, both of which have a similar impact on the total portfolio (as you will see shortly).

First, you might cut the allocation to Treasuries to add other asset classes. Some investors have tried to replace Treasuries with bonds of lower credit quality to increase yield and decrease exposure to what they view as inevitably rising interest rates. The problem with this approach is that a bond with a lower credit quality has a different economic bias than Treasuries. This makes sense because Treasuries have virtually no credit risk, while bonds of other issuers do have varying degrees of risk of default. During weak economic periods the general risk of default rises and therefore adversely impacts lower quality bonds. On the other hand, Treasuries thrive in such an environment because of the flight to quality. In this way, Treasuries are unique securities and there are very few, if any, true substitutes (within the United States at least; of course you can always buy high quality sovereign bonds overseas).

A second way to reduce exposure to Treasuries is to shorten the duration of your Treasury holdings. The thinking is that the shorter the duration, the less interest-rate sensitive the bonds. And since interest rates are expected to rise, reducing duration seems like a reasonable reaction to this problem. It is true that shortening the duration of the Treasury holdings reduces volatility, but more importantly, it also negatively impacts the effectiveness of these holdings. What if rates actually fall instead of rise, or if they don't rise as much as everyone expects? What economic environments would those outcomes likely involve, and what would be the consequences to your portfolio? These are crucial questions that ought to be part of the thought process. Table 5.1 summarizes the changes investors have made in response to concerns about low interest rates and their corresponding impacts on the portfolio.

We can test the conceptual linkages described in Table 5.1 by comparing the actual returns of the various substitutes for Treasuries during the period from October 2007 (when interest rates first fell to theoretically low levels) to December 2013 (when this book was written). By analyzing how each conventional solution to low interest rates behaved during various environments, we can gain insight into the diversification attributes of each asset class. When you look at the data, try to focus on how each market segment performed when the stock market was going up or down: How did Treasuries do when equities were performing well and how did they do when equities experienced declines? How did shorter-term bonds and lower quality fixed income perform during similar environments? The correlation between asset classes is a critical part of building a balanced portfolio, so the intent here is to demonstrate performance using recent examples of the types of relationships that you might expect.

The results are presented in Table 5.2. The table divides the time frame since 2007 into bull and bear market phases of the stock market, which mostly correspond to strong and weak economic growth cycles, respectively (since inflation has been relatively stable during this period). In addition to long-term Treasuries, the returns of three other asset classes are compared to equity returns during each cycle. Intermediate bonds and cash represent the strategy of reducing bond market duration to protect against rising interest rates. High-yield bonds offer a higher yield and theoretically better downside protection against rising interest rates.

Source: Bloomberg.

All returns cumulative.

Indexes used are as follows: equities (S&P 500), long treasuries (Citigroup US 30 Year Treasury), intermediate bonds (Barclays Aggregate Bond), cash (3 month Treasury repo rate), and high yield (US High Yield Master II).

Specific time periods used are as follows: 10/9/07–3/9/09, 3/9/09–4/23/10, 4/23/10–7/2/10, 7/2/10–4/29/11, 4/29/11–10/3/11, 10/3/11–4/2/12, 4/2/12–6/1/12, 6/1/12–12/31/13.

These represent peak to trough and trough to peak periods for the S&P 500 Index since October 2007.

None of the popular substitutes for Treasuries offered similar upside during periods that equities suffered losses. The critical lesson to draw from the analysis in Table 5.2 is an appreciation of why the return patterns of the recent past materialized. The economy underperformed expectations and therefore long-term Treasuries, which are biased to outperform in such periods, soared in value. It certainly could have gone the other way. In fact, there were periods during which Treasuries did poorly during this great run. Of course, equity gains during those times more than offset the underperformance in Treasuries.

Clearly, of all the bond substitutes—intermediate-term diversified bonds, cash (both of which are proxies for shorter-duration bonds), and high-yield fixed income—Treasuries proved to be the best diversifier. This analysis has nothing to do with the fact that Treasuries outperformed most of the rest (high-yield bonds ended with better absolute performance, although the pattern of returns generally matched those of equities). Indeed, the opposite result would have yielded the same conclusion. The returns shown in Table 5.2 should be expected because of the environmental bias of this crucial asset class. Intermediate diversified bonds, although they are of high quality, do not have sufficient volatility to materially help the portfolio during economic weakness. Cash and low-duration fixed income strategies don't move enough to make a difference. High-yield bonds, which are more credit-sensitive than interest rate-sensitive, predictably tend to correlate with the ups and downs of the stock market. All three commonly used bond alternatives offer inferior diversification benefits when compared to long-term Treasuries.

The three major oversights that I just covered help demonstrate the flaws in conventional thinking as it relates to the Treasury asset class. In the rest of this chapter I reintroduce Treasuries using a balanced portfolio perspective. As you will read, the focus will not be on the future expected return of Treasuries as is the conventional approach.

Economic Bias: Treasuries Favor Falling Growth

Weak growth benefits Treasuries because of the increasing likelihood of falling interest rates. This occurs because a weakening economic environment produces a reaction by the central bank to consider lowering interest rates to encourage borrowing to spur economic growth. When growth unexpectedly declines, it results in a move by the Fed that had not been anticipated by the market since the falling growth itself had not been expected. As a consequence, interest rates decline because cash rates suddenly are discounted to rise less than previously anticipated.

Declining interest rates are great for Treasuries because the bonds you own have a fixed interest rate. If prevailing rates are lower than those discounted at the time of your purchase, then the higher coupon of your bond relative to market rates makes your bond a more attractive investment. Consequently, the price rises to reflect the new reality of lower interest rates as this gradually becomes apparent to the market.

Moreover, Treasuries tend to do well during weakening growth periods because of a flight to quality during adverse economic environments. When bad things happen to the economy, it seems reasonable that investors would react by becoming more cautious and risk averse. Treasuries, which have no credit risk, all of a sudden become more attractive investments because the focus shifts from maximizing returns to protecting capital. The high-quality bias of Treasuries attracts frightened investors who are seeking a safe place for their money.

Within this context you might better appreciate how low interest rates can go even lower. All it takes is for economic growth to underperform what is already discounted. Since growth underperforms expectations roughly half the time from a historical standpoint, it makes sense that the odds of low rates dropping are similar to the odds of them rising (even with low starting yields).

Economic Bias: Falling Inflation Benefit to Treasuries

Lower inflation also benefits bond prices. As I expressed in the opening chapter, the Fed attempts to keep inflation at bay by raising interest rates to deter borrowing when inflation is too high and lowering rates to stimulate debt growth when inflation is too low. Each incremental change in inflation produces an expectation of its future path. How that expectation relates to what has already been discounted in the original price is then reflected in the current price. If inflation falls more than expected, then that leads to a greater likelihood of the Fed lowering short-term interest rates. This new discovery then feeds into the increased probability of lower future long-term interest rates, which benefits Treasury prices.

Putting It Together: Falling Growth and Falling Inflation (and Deflation) Economic Bias

As described above, lower interest rates benefit bond prices. More precisely, it is the future path of cash rates that influences bond prices. This is because the bond yield at various maturities from short-term to long-term merely reflects the discounted changes in cash rates. Since the Fed controls cash rates and the Fed responds to changes in the economic environment (in terms of growth and inflation) by adjusting cash rates, then these economic shifts together impact discounted cash rates. Changes in cash rates then impact longer-term interest rates, which directly affect Treasury bond prices.

Table 5.3 provides historical data that supports these cause-effect linkages. The rare deflationary periods that we've experienced have been bundled with falling inflation economic climates in this table.

As you might expect, at 1.4 percent the historical average excess return above cash is not very high. Treasuries have the same credit risk as cash, so investors in long-term Treasuries only receive compensation for the interest rate risk that is taken. As is the case with equities, the difference in returns by economic environment is meaningful. During falling growth periods the excess return of Treasuries has averaged 5.5 percent per year. This is a very strong average return that is nearly identical to the 5.6 percent equity average excess return (for all periods). Considering that about half the time growth is falling relative to expectations, it is easy to see the benefits of owning Treasuries as part of a balanced portfolio.

Of course, the opposite environment of rising growth has historically been very bad for Treasuries, with an average excess return of –3.0 percent. Predictably, Treasuries benefit from falling inflation and underperform during periods of rising inflation, as demonstrated in Table 5.3.

Figure 5.2 provides similar data for longer-term rolling 10-year periods. Notice that the good times occurred during predictable economic climates, as did the bad environments.

The Benefits of Higher-Volatility Treasuries

One very important and somewhat counterintuitive point about the returns in Table 5.3 needs to be reinforced. The only reason the returns diverge materially between the good and bad periods is because we are using long-duration Treasuries. Shorter-duration Treasuries would share exactly the same environmental bias as longer-term Treasuries. In fact, the excess returns above cash are nearly identical. However, the returns during good environments and bad environments would not be too different. Good periods would outperform the average excess return by a little bit and bad environments would produce slight underperformance. This is simply because longer-duration bonds have more volatility than short-term bonds. More volatility means that the return pattern will fluctuate more, which leads to the wider divergence in returns during good and bad economic climates.

From a conventional perspective, when you have two asset classes that produce the same returns you would always select the one that has less risk or lower volatility. Why get the same return for more risk? However, from a balanced portfolio viewpoint, you do not make asset allocation decisions based on the same criteria. Instead the emphasis is on identifying the economic bias of each asset class and determining its volatility. As is apparent from Table 5.3, the reason more volatility is preferred over less volatility (and therefore long-duration instead of short-duration Treasuries) is because of the better returns when you need them. This is the whole idea behind building balance. You focus less on each individual asset class and more on how it fits within the bigger picture. Each asset class is part of a team and serves a bigger purpose.