by machinehead
Many of you will recognize today’s author from his insightful comments that appear frequently across PeakProsperity.com.
It sucks to try earning income from investments these days. Until about ten years ago, most folks assumed they could make an easy 5 percent from safe, risk-free vehicles such as T-bills or CDs. With $500,000 saved, you could generate $25,000 in annual income. Them days are gone! Today, thanks to the Federal Reserve’s Japanese-style ZIRP (Zero Interest Rate Policy) regime, one-year T-bills yield only 0.25%, while one-year CDs average 1.25% — a mere $6,250 annually on a $500,000 account.
Three years ago, I confronted a related aspect of this problem while serving on the board of a church which needed income. In the summer of 2007, it received a $500,000 bequest. The headquarters organization recommended putting the entire sum into their balanced fund — meaning 60% stocks, 40% bonds. The Wall Street manager’s sales pitch claimed that since market timing is ‘impossible,’ ‘now’ would be a good time to buy into the fund — always had been, always would be. No thinking required!
But I vehemently objected to this notion, particularly in regard to the stock portion of the fund. In the summer of 2007, the dividend yield on the S&P 500 stock index was under 2% — not only in the danger zone of overvaluation (more on this later), but also resulting in a sub-1% yield after management fees. As we now know, stocks proceeded to lose over half their value from October 2007 to March 2009.
Fortunately, we reached a compromise: the bequest would be invested in four annual installments, instead of all at once. The first 25% installment, at the end of 2007, of course lost badly in 2008. However, the result vividly illustrated the merit of holding two asset classes instead of one. While the 60% stock portion of the fund lost 37%, the 40% bond portion gained 3% as interest rates sank. The fund’s net result was a 21% loss in 2008 — on one-fourth of our portfolio. Since the rest of the bequest was safely stashed in short-term T-notes (which paid a richer 3.8% yield) outside the fund, the overall portfolio was barely dented.
Today, with much lower yields on offer, it’s still possible to obtain a reasonable income. Unavoidably, this means taking price risk. However, by diversifying among asset classes, some of the price risk can be hedged. As an example, in an idealized business cycle, bond prices peak first, followed by stocks, then by commodity prices. The most recent cycle conformed to the pattern: bond prices topped in June 2003; stocks in October 2007; and commodities in July 2008 (seven months after a recession began). Since each asset class follows its own rhythm, one is often rising while another is falling, thus damping the fluctuations of a combined portfolio.
Worldwide, there are three major asset classes:
- Fixed income — money lent at interest.
- Stocks — equity investment in publicly-traded and privately-held enterprises.
- Property — real estate, commodities, collectibles and intangible property.
According to Albert J. Brenner at aametrics.com, the global bond market was valued at $82 trillion and global equities at $44 trillion in late 2009. [Ref. 1] Global property is probably in the $80 trillion ballpark as well, since The Economist had estimated its value at $62 trillion in developed countries alone in 2002. In total, these Big Three asset classes are valued at about 3.5 times global GDP of $60 trillion.
But are they cheap or expensive? As a first glance, let’s look a selection of current yields in the US — the cash income these assets offer:
- Fixed income: 10-year Treasury notes — 2.8%
- Stocks: Dow Jones Industrial Average — 2.9%
- Real estate: average of four REIT ETFs* — 3.2%
* REIT (Real Estate Investment Trust) ETFs (Exchange Traded Funds): symbols ICF, IYR, RWR, VNQ
Right off the bat, it’s obvious that nothing is cheap — these are historically low levels of income from all three sources. However, the 10-year T-notes are the most extreme of the three. Having yielded a high of over 15% in 1981 and a low of just over 2% in 2008, T-notes are at the bottom end of their yield range. Treasury prices are correspondingly high, and slide when yields rise. (TLT, a long Treasury bond fund, dropped 5.5% in the first 10 days of September on a slight rate hiccup.)
As for US stocks, over the past century their average dividend yield was around 4%, with 3% formerly considered to be the threshold of overvaluation when the yield fell below this level. But since the mid-1990s, a dividend yield under 2% has served more effectively as a marker of overvalued stocks.
In the case of commercial real estate, capitalization rates (net operating income divided by price) of 7 to 10% are typical. According to Hai-yong Yu at Bloomberg, cap rates averaged 7.22 percent in the second quarter — a little on the low side, yet 429 basis points, or 4.29 percentage points, higher than the yield on 10-year T-notes and near the record spread of 539 basis points in the first quarter of 2009. [Ref. 2] In other words, the real estate cap rate is attractive compared to low yields on T-notes and stocks. REIT ETFs pay dividends well below the cap rate, but they still beat the yield from stocks and bonds.
What can we expect in the future from these asset classes — say, over the next ten years? For fixed income, instead of Treasuries only, let’s focus on a broad index which includes all investment grade bonds, both government and corporate: the Barclays Aggregate Index. An ETF which tracks this index, AGG, yields 3.33%. If we make the charitable assumption that yields stay at their current low level, 3.33% is all that we can expect for future total return. If yields rise, total returns would be even lower for awhile, since rising yields erode the prices of already-issued bonds.
To get a sense of the risk in bonds, note that AGG’s price has fallen 1% since its August 31st high — equal to almost four months of interest income lost. If AGG’s price fell back to the level of early April, it would equal a loss of 18 months worth of interest. AGG’s bonds are only about one-third as volatile as stocks. But unlike T-bills and CDs, which carry no price risk when held to maturity, bond funds may experience occasional flat or losing years, especially when their yields are low.
Using a composite of valuation models, Dr. John Hussman of Hussman Funds has estimated [Ref. 3] that US stocks are currently priced to deliver average total returns of about 6% over the next decade — with plenty of volatility: typically 16% annually. This means that even if stocks do deliver a compounded total return of 6%, one would expect three or four losing years per decade, with one or two of them being harsh double-digit percentage losses. Nevertheless, over many decades, stocks have delivered higher total returns than bonds, which is why pension funds continue to own them.
Like stocks, real estate historically has delivered about half of its total return in the form of price appreciation. While officially-measured inflation is currently low, the Federal Reserve radically expanded its balance sheet in 2008-9, and in August 2010 backed away from promises to let it shrink as mortgage securities are paid off. My belief is that the Fed will never be able to let its asset purchases run off, and that this large expansion of the monetary base will prove inflationary later on. Even modest 3.8% inflation (the annually compounded average since 1946), augmenting both property prices and rents, would produce total returns of 7% for REITs.
Thus, our ranking of expected returns is consistent with the earlier comparison of current yields. The lowest yielding asset class, bonds, also has the lowest projected returns, while REITs garner the top honors in both comparisons. This isn’t surprising, since a low current yield directly reduces future return prospects, and vice versa.
Now let’s make a final rough-cut comparison — of popularity. Unscientific, but from a contrarian point of view, we want to favor unpopular assets, while underweighting those enjoying too much enthusiasm. Here is my idiosyncratic poll:
- Fixed income — POPULAR: unprecedented, Bubble-like inflows are going into bond mutual funds.
- Stocks — DOUBTED: outflows from stock mutual funds continue even after gains since early 2009.
- Real estate — HATED: daily, harshly bearish articles bemoan the dreadful state of housing.
Although broad-brush and impressionistic, this informal popularity ranking confirms the other two rankings. We need to be cautious about bonds, while giving real estate at least equal billing to stocks.
Building a robust multi-market portfolio
In constructing a portfolio to implement these insights, let’s start with the 60/40 balanced portfolio mentioned at the beginning. The idea of a portfolio balanced between stocks and bonds goes back at least to 1935, when Gerald Loeb published The Battle for Investment Survival. It remains a core principle of pension fund management to this day.
As we’ve seen this decade, stocks can be very bad actors during secular bear markets, because of their extreme price swings. With economic prospects dim and dividend yields still low, it’s not at all clear that this secular bear market is over. Meanwhile, as discussed, slightly higher returns are expected from real estate than from stocks.
Therefore, instead of devoting 60% of the portfolio to stocks, we’re going to split it between stocks and REIT ETFs, giving them 30% each in the overall portfolio. Moreover, the stock portion is going into high-dividend funds — not only for higher income, but also because research shows that there is no penalty to be paid in price appreciation. In other words, applying an income-oriented value screen to stocks produces excess return compared to the broad, unfiltered universe of stocks. [See Arnott, Hsu and Moore, Fundamental Indexation, Ref. 4.] Furthermore, since economic growth in emerging markets is trouncing that of developed markets by 4%, we want half of our stock allocation overseas. So, here are the proposed stock fund selections and portfolio weights:
- 15% DVY — iShares Dow Jones Select Dividend Index. Yield: 3.70%
- 15% DEM — Wisdom Tree Emerging Markets Equity Income Index. Yield: 5.74%
As for REIT (Real Estate Investment Trust) ETFs, let’s select two large ones:
- 15% ICF — iShares Cohen & Steers Realty Majors Index Fund. Yield: 2.88%
- 15% VNQ — Vanguard REIT ETF. Yield: 3.56%
Now let’s consider the 40% of a balanced portfolio traditionally devoted to bonds. Our analysis suggests that bonds are unattractive — long maturity bonds have price risk, while short-term bills offer only fractional yields. One way to cut price risk while retaining some yield is in intermediate term bonds. So we’ll put 10% of the portfolio into CIU, whose effective duration is 4.24 years. This means that a 100 basis point (one percentage point) rise in prevailing interest rates would clip 4.24% from the value of the fund’s bond portfolio. By contrast, the duration of the 10-year T-note is 8.11 years; a 100 basis point rate rise would cost it a punishing 8.11% price decline. The proposed ETF:
- 10% CIU — iShares Barclay Intermediate Credit Bond ETF. Yield: 3.99%
Since bonds are unattractive, let’s consider an alternative source of income: master limited partnerships (MLPs), which derive their fairly stable high incomes from midstream energy services, primarily pipelines and storage. A diversified portfolio of MLPs is available in ETNs (Exchange Traded Notes), such as AMJ. However, most ETNs have a management fee of 0.85%, which I regard as undesirably high — not to mention that ETNs are a debt obligation of their issuers. So for MLPs, we will depart from our use of diversified ETFs elsewhere, and select two individual MLPs:
- 10% KMP — Kinder Morgan Energy Partners LP. Yield: 6.35%
- 10% EPD — Enterprise Product Partners LP. Yield: 6.04%
Finally, although short-selling bonds in order to profit from a price decline is mathematically unattractive (owing to the required payment of the interest coupon to the lender), we can still include an asset which is negatively correlated to bonds: gold. Although gold stocks do not produce much income, they do serve as a potential hedge against financial calamities ranging from sovereign defaults to weakness in the US dollar. Since this portfolio is denominated in US dollars, we’ll dedicate 10% of it to a golden insurance policy:
- 10% GDX — Market Vectors Gold Miners ETF. Yield: 0.21%
In total, we’ve constructed a portfolio diversified among five different markets: stocks; REITs; bonds; energy MLPs; and gold. Call it the 5-market model (5MM) — not to be confused with 5.56 mm ammo! Yet it uses only eight positions, in order to minimize management and commissions. Its overall yield is a respectable 4%. No one trading symbol constitutes more than 15% of the portfolio. And each fund itself, other than the two MLPs, holds dozens of underlying issues, so that the failure of any single entity would barely affect the portfolio.
Summing up 5MM’s allocations by category, they are:
- 30% stocks
- 30% REITs
- 20% energy MLPs
- 10% intermediate bonds
- 10% gold stocks
Another way to look at our 5MM portfolio is to refer back to the original 60/40 balanced fund used as a starting point. Now our allocations to these two traditional asset classes are down to only 40% of the entire portfolio: 30% in stocks, 10% in bonds. Both these allocations are tweaked: half the stocks are US high-dividend; the other half are emerging markets high-dividend. The bonds are of intermediate duration to reduce price risk. Moreover, as explained below, the stock/bond allocation can shift between 30/10, 20/20 and 10/30 depending on stocks’ valuation.
Fully 60% of the 5MM portfolio is in what would be called alternative investments from the traditional Wall Street point of view: 30% REITs; 20% energy MLPs, 10% gold stocks. Not only does this portion of the portfolio offer a solid yield, but also it has a distinct property and commodity-oriented flavor. These asset classes offer inflation protection, but also have their own price cycles, distinct from those of stocks and bonds. This reduces the chance that all categories of the portfolio would experience price declines at the same time, and measurably reduces risk in the form of price volatility.
Maintaining and modifying the 5MM portfolio
First off, please consider the 5MM portfolio as a baseline for your own thinking. If you’re averse to having 30% in stocks or 10% in gold, then reduce these categories and increase another, such as intermediate bonds. Raising the lower-risk bond allocation cuts overall risk and volatility of the portfolio. Gold stocks are the highest volatility member of the group, but serve an insurance function that the other categories don’t. (Disclosure: GDX is the only issue in the 5MM portfolio that I own.) However, there’s another way of cutting risk discussed below.
Other sponsors offer ETFs with similar investment objectives to those mentioned here, which could be substituted. Pay attention to their market capitalization, management fees and daily trading volume: some specialized ETFs are small-cap, costly in fees and thinly traded.
Because of its diversified holdings, the 5MM portfolio could be held for years without changes other than annual rebalancing, to bring the proportions of each holding back into line with their targets. Annual rebalancing is done when a component gets more than a tenth out of line with its target weight. For instance, if a 15% component ends the year valued at over 16.5% of the portfolio, that position would be sold down to 15%, and the proceeds reinvested in underweight components.
A second annual structural adjustment, when indicated, is reducing the two stock fund allocations (DVY, DEM) from 15% to 10% each if the S&P 500 dividend yield is less than 2.0% at the time of annual evaluation (as it was in Dec. 2007, for instance); or to 5% each if the S&P 500 dividend yield is under 1.5% at the end of the period (as it was in Dec. 2000, to mention another pre-recession reading). Under these low-dividend, overvalued-stocks conditions, the allocation to intermediate bonds (CIU) is increased to 20% or 30%, respectively, as the stock weightings are reduced.
With the S&P 500 dividend yield currently hovering at the 2.0% level, the first de-weighting criterion might come into play when the portfolio is next evaluated at the end of 2010. Since the dividend yield is reported with a lag, use the December 1st monthly value for the end-of-year evaluation. You can track the monthly S&P 500 dividend yield at multpl.com, the site listed in Ref. 5.
A third structural adjustment recommended for retirees is to halve risk by placing half of their portfolio in one-year CDs (the longest maturity I would recommend, since rates likely will rise in the future), and the other half into the 5MM portfolio. We’ll call this the 5MM/2 portfolio.
It takes an unusual combination of correlated drops across asset classes (such as occurred in 2008) to cause a double-digit annual loss in the 5MM/2 portfolio. Yet it still delivers a blended 2.75% current yield (assuming 1.5% yield on the CD portion), and is expected to add two or three percentage points in annual price gains over a holding period of ten years. If a 3.0% current yield is your non-negotiable minimum, a 40% CDs / 60% 5MM split will deliver it. Or, if you want to cut risk even further, increase the CD allocation to above 50%. Any level of risk tolerance can be catered for in this manner.
Performance simulation: 5MM and 5MM/2
Let’s look at a rough estimate of how the 5MM and 5MM/2 portfolios would have performed from the end of 1999 through the end of August 2010, a period of 10.67 years. This was the worst decade for stocks since the 1930s.
Of the eight issues in the 5MM portfolio, only the two energy MLPs existed a decade ago. ETFs, a new idea in the 1990s, gained widespread acceptance in the early 2000s, and many new ones were created. For the six ETFs, performance before their issuance was simulated by substituting older ETFs (SPY for DVY, EEM for DEM, and AGG for CIU) and, before the older ETFs’ existence, using indexes such as the MSCI Emerging Markets index, the NAREIT index of REITs, the Barclays Aggregate Index of bonds, and the XAU gold and silver stock index. No allowance was made for commissions and slippage at annual rebalancings, or for taxes.
Because of unavoidable errors introduced by substitutions, the simulated performance figures must be regarded as approximate, impressionistic and optimistic. As usual, past performance is no guarantee of future results. But an important added caveat applies: the preceding disclaimer is standard for actual portfolio past performance data. A simulated portfolio introduces the serious additional distortion of hindsight bias. Let’s not beat around the bush about this head-fake headache.
Here is one example of hindsight bias: in constructing this portfolio, I knew that REITs, energy MLPs, emerging market stocks and gold stocks rose strongly during the past decade. Although I was extremely bearish on stocks at the time, had I constructed this multi-market portfolio at the end of 1999, I might not have included the same alternative asset classes, and perhaps not at the same weightings.
A second example of hindsight bias: although the dividend yield thresholds which dictate de-weighting stocks are simple round numbers (2.0% and 1.5%), it’s likely in 1999 that I might have used the traditional 3% criterion, or some other levels less finely tuned to the known events of the past decade.
One kind of bias which is not present is optimization — that is, iteratively varying the component weightings to tease out the highest return. The 5MM portfolio was structured and described before the performance testing was done. No subsequent changes were made to boost the results.
It’s axiomatic that a portfolio selected with hindsight bias will not perform as well going forward as it did in the known, certain past. The shortfall might be two, three, even five percentage points of compounded annual return. Only robust underlying design principles, such as asset class diversification and simple rules for portfolio adjustments, can minimize the inevitable performance degradation as markets embark on new secular trends.
Now, with these critical caveats in mind, let’s compare the baseline 60/40 balanced portfolio (60/40 BP); the 5MM portfolio; and its diluted version with a 50% anchor in CDs — 5MM/2. Values in the table are the total return in each year, expressed as a percentage:
Year |
Stocks/bonds (60/40 BP) |
5-mrkt model (5MM) |
50% CD + (5MM/2) |
2000 |
-1.32 |
21.13 |
14.11 |
2001 |
-4.09 |
16.85 |
10.72 |
2002 |
-9.22 |
5.55 |
4.26 |
2003 |
18.40 |
33.39 |
17.72 |
2004 |
8.08 |
14.83 |
8.75 |
2005 |
3.83 |
12.35 |
8.27 |
2006 |
10.70 |
19.53 |
12.47 |
2007 |
5.71 |
1.59 |
3.52 |
2008 |
-20.75 |
-25.75 |
-10.91 |
2009 |
19.79 |
34.14 |
18.03 |
2010* |
-0.91 |
11.76 |
6.38 |
* through August 31st.
Here are summary statistics for the three portfolios:
Results |
Stocks/bonds (60/40 BP) |
5-mrkt model (5MM) |
50% CD + (5MM/2) |
Compounded annual return |
2.16% |
12.39% |
8.46% |
Standard deviation |
12.20% |
16.52% |
8.14% |
Sharpe ratio |
-0.03 |
0.60 |
0.73 |
$100,000 grows to … |
$125,657 |
$347,527 |
$237,776 |
To begin with, the compounded annual return (CAR) from holding 3-month T-bills over the 10.67 year period — the so-called risk-free rate — was 2.59%. In T-bills, there’s never a losing year — only a steady progression in equity. However, the CAR from T-bills was almost exactly equal to the compounded annual change in the CPI (Consumer Price Index) over the period — 2.55%. In T-bills, one only kept even with inflation, not gaining any purchasing power from the interest paid.
In regard to beating inflation, stocks performed superbly in the second half of the 20th century. A typically optimistic late 20th century analysis, cited at Wikipedia, assumed that stocks offer an average 10% total return, while T-bills return about 3.5% annually [Ref. 6]. But stocks’ 6.5%