Thursday, August 14, 2014

It's All in the Math - Equity Returns over the Coming Decade (2 of 2)

While equity markets have more "moving pieces" than their fixed income counterparts, their longer-term outcomes can likewise be broken down into a handful of understandable and forecastable components. When approached in this disciplined manner, we see that returns from nearly every major equity region in the coming decade will be poor, with the UK faring better than most and the US likely to struggle.  

It is a given that there are only two ways to make money from an equity - either you are paid a stream of dividends, or the price another party is willing to pay you for your shares differs from the price you paid (Dividend + Price Change).  In general, the Dividend component is readily observable and tends to be less volatile while stock Price Changes are more volatile and, in the short-run, more unpredictable.

To better understand the portion of return due to Price Change, we can further decompose that piece into Earnings Growth per share and the P/E Ratio (i.e. the price one is willing to pay for each dollar's worth of those earnings).  Both are somewhat volatile, but less so than the overall Price Change itself.  To simplify further, I sub-divide the Earnings Growth component into Revenue Growth and the change in the profitability of those Revenues (i.e. the Profit Margin).  Taking all this together then, if we know with near certainty just four variables: Dividends, Revenue Growth, Profit Margins and the P/E Ratio, we know with certainty the return from an equity (or equity index).

Fortunately, for the purpose of our analysis, two of these components are relatively stable over a period as long as a decade (Dividend Yield and Revenue Growth), while valuation and profitability are more volatile -- and it is this change in the P/E Ratio and Margins that ultimately drives returns.  Thus a sensitivity analysis featuring these two unknowns is called for.
Some Background
Equity earnings tend to be cyclical over time (see chart) and gauging market value on a single, spot observation can be mis-leading -- especially when earnings are well above, or well below, their trend. As of June 30th, the P/E Ratio on trailing, reported earnings of the S&P 500 index was 18.2x, just above the longer-term norm of 17.1x (the median valuation since Dec 1959).  One might therefore conclude that the market is slightly overvalued. What is left out of such a simplistic analysis is that the earnings part of that calculation is not at its norm, but rather closer to a cyclical high.

Instances such as this occur when margins are historically high, even as they are understood to be generally mean-reverting.  Today, margins in the US and Japan are closer to one of these cyclical peaks, while those in the UK and Europe are not.
When valuation is compared to reported earnings per share, and the cycle is not taken into account, valuation (as measured by the PE Ratio) tends to be more volatile.  When the index price level is measured against trend EPS (that is, assuming a constant "normalized" margin), valuation extremes tend to disappear and a more stable relationship is observed.
Thus, a valuation analysis which adjusts for the margin's deviation from its norm and which assumes dividend growth in line with earnings (as long as the current payout ratio is near longer-term norms), revenue growth in line with historical trends (typically that of nominal GDP) and a valuation component (P/E ratio) that returns to its long-run median over a period of ten-years should provide a decent assessment of where returns are headed as cycles run their course.  Those who disagree with my assumptions of a return to normal margin can use the accompanying sensitivity tables and observe returns under a range of margin and P/E inputs - but refer to the above charts to understand what is in the "normal" (i.e. likely) range of outcomes.  I would direct doubters back to "The Profitability Illusion" (posted to this blog June 15, 2013) which discusses in some detail why over 2 percentage points of the current S&P 500 margin is illusory and unlikely to be sustained (all due to lower interest expense as opposed to better operational efficiency).  In any case, the range of returns can be illuminating and the dispersion of returns, not as wide as one might assume.

I turn first to the United States and use the S&P 500 index as a proxy for US Equities.  Over the past 50+ years, a median valuation of 17.1x earnings is the norm (I choose to use a median as opposed to an average because out-sized values can distort its calculation.  In full disclosure, the average P/E has been 17.6x).  Margins have averaged 6% for the past 50 years and have proven throughout to be mean reverting.  Today, though they stand near record highs of over 8% and in this analysis it is assumed they will slowly return to the 6% average.
For this exercise, we also assume 5.5% annual compounded nominal revenue growth (in line with historic norms) and a 2.08% annual return from dividends (normalized current yield). Putting this all together, our best guess return estimate for US Equities over the next 10 years is 2.0% per year.  If you want to assume record margins continue, you could up that expectation, but just to 5.0% (see table below). Under no reasonable scenario, are double digit equity returns in sight.

S&P 500 10-Year Return Estimation (5.5% Nominal Revenue Growth, 2.08% Annual Dividend, Various Margin and P/E Scenarios
    Normalized PE Ratio
  2.0% 13 15 17 19 21
Normalized Margin 4.0% -4.5% -3.1% -1.9% -0.9% 0.1%
5.0% -2.4% -1.0% 0.2% 1.3% 2.3%
6.0% -0.6% 0.8% 2.0% 3.2% 4.2%
7.0% 0.9% 2.3% 3.6% 4.7% 5.8%
8.0% 2.2% 3.7% 5.0% 6.1% 7.2%
Taking Japan next we conduct a similar analysis, using the historically lower 3.0% margin levels and 3.0% nominal revenue growth.  Though Japanese valuation data is distorted somewhat by 10+ years of a great bubble, we feel a normalized P/E ratio of 20.x, higher than that of the US, can be argued, though we show a range of 12.5x to 27.5x in the sensitivity below.  Still, a 3.0% return is the most likely annual outcome for the coming decade.
MSCI Japan 10-Year Return Estimation (3.0% Nominal Revenue Growth, 1.77% Annual Dividend, Various Margin and P/E Scenarios
    Normalized PE Ratio
  3.0%            12.5            15.0            20.0            25.0            27.5
Normalized Margin 2.0% -5.4% -3.7% -1.0% 1.2% 2.2%
2.5% -3.3% -1.6% 1.2% 3.5% 4.4%
3.0% -1.6% 0.2% 3.0% 5.3% 6.3%
3.5% -0.1% 1.7% 4.6% 6.9% 7.9%
4.0% 1.2% 3.0% 6.0% 8.3% 9.4%

Moving on to the European markets and the UK, we reach a somewhat happier, though still below historic, level of return expectations.  For the UK market we use similar revenue (5.5%) and normalized margin assumptions (6.0%) as we do for the US.  For the continent, I discount both revenue growth and margins by 0.5%, in line with historic observations.  Again, under a broad range of valuation and margin assumptions we see a most likely return for the UK of 5.7% and a most likely return of 3.9% for the continent.
MSCI UK 10-Year Return Estimation (5.5% Nominal Revenue Growth, 3.95% Annual Dividend, Various Margin and P/E Scenarios)
    Normalized PE Ratio
  5.5% 10 12 14 16 18
Normalized Margin 4.0% -1.7% 0.1% 1.6% 2.9% 4.1%
5.0% 0.5% 2.3% 3.8% 5.2% 6.4%
6.0% 2.3% 4.1% 5.7% 7.1% 8.3%
7.0% 3.8% 5.7% 7.3% 8.7% 10.0%
8.0% 5.2% 7.1% 8.7% 10.1% 11.4%
MSCI Europe ex UK 10-Year Return Estimation (5.0% Nominal Revenue Growth, 3.71% Annual Dividend, Various Margin and P/E Scenarios) 
    Normalized PE Ratio
  4.0% 10 12 14 16 18
Normalized Margin 3.5% -3.8% -2.1% -0.6% 0.7% 1.9%
4.5% -1.4% 0.3% 1.9% 3.2% 4.4%
5.5% 0.5% 2.3% 3.9% 5.3% 6.5%
6.5% 2.2% 4.0% 5.6% 7.0% 8.2%
7.5% 3.6% 5.5% 7.1% 8.5% 9.7%
It is almost certain that returns from the major developed equity markets are quite likely to remain in single digits - from somewhere near 2.0% compounded in the US to under 6% compounded in the UK – for the decade ahead. This realization, combined with similar fixed income expectations is that most plan sponsors will have great difficulty achieving their projected return assumptions of over 7.5%, even under the most optimistic of market conditions
Obviously a new perspective on managing and allocating funds is called for. This is why my July blog post, “The Task of Capital Allocators” is so important for every Plan Sponsor to consider.


Monday, August 4, 2014

It's All in the Math - Fixed Income Returns for the Decade Ahead (1 of 2)

It is becoming a more common belief that equity and bond market returns are likely to be lower than normal in the years ahead.  If proven out, this will have negative implications for the many retirement plans that have target return assumptions based off of historical norms.  Shortfalls in returns will require additional contributions - either out of corporate profits (for private plans), or out of the pockets of taxpayers (for public plans).  Expected blended return expectations remain above 7.5% for most plans.  Returns from traditional fixed income and equity markets will not reach this level and a rethinking of expectations and asset allocations will be necessary. 

Using models I originally developed in the late 1990’s, we can arrive at accurate return projections for both fixed income and global equity market over the coming decade. 

Starting with fixed income markets.  A bond is a fairly straight forward instrument with returns accruing to just four factors: the price paid, the coupon payments received, the reinvestment of those coupons and the ultimate return of principal.  Taking each in turn, we know the price paid.  We also know the coupon payments as they are contractual in nature.  The reinvestment return of these coupons is unknown; however, as I will demonstrate, even an immediate, radical move in interest rates will not dramatically change the overall return of a bond over its lifetime.  Finally, while return of principal to a single bond may be uncertain, when looking at the broader market of bonds of similar ratings, historical experience can provide a reasonable guide as to default and recovery rates.  Putting these together, estimating long-term bond returns is a very straight forward process.

At July 30th, you could buy a ten-year US government note with a maturity of May 15, 2024 for a price of $99.49.  That note will pay a semi-annual coupon at an annual rate of 2.50%.  At maturity, an investor will receive $100 and along the way, twice yearly coupon payments of $1.25 (per $100 value).  While we do not know the rate at which those coupons will be reinvested, even if we assume rates rise by 500 basis points before the first coupon payment is received (to 7.56% across all maturities), the total annualized return from this note will rise only to 2.85% over the ten year period.  Conversely, if rates fell to zero and an investor received no return at all on the coupons, the ten year total annualized return on this note falls only to 2.29%.  Thus, assuming no default, one can not expect anything other than a return of between 2.29% and 2.85% from buying a ten-year US government note today.  Another way of looking at this is, assuming no risk of default, the best approximation of the long-term return on a bond is probably just its current yield (on the UST, currently 2.56%).  Of course, the road to that compounded annual return need not be smooth.  Outsized returns or losses in a near-year, will cause out-year returns in the other direction as the result is mathematically determined.

Looking at other developed government bond markets (G-7 plus Australia and Spain), we note 10-year yields between 0.53% (Japan) and 3.43% (Australia).  Assuming no risk of default, these again are the best estimate for annualized local currency returns over the coming decade. 

Source: Bloomberg

Of course, given the recent experience of certain governments (Argentina, Greece), zero chance of default might not be the best assumption.  If one wanted to insure themselves against default risk, we can look to the Credit Default Swap market to gauge the costs of protection.  Higher yielding markets such as Spain and Italy currently have annual “insurance” costs of 1.15% and 1.56%, respectively.  Perceived “safe” markets like the US and Germany have lower insurance costs of 0.25% and 0.42%, respectively.  Calculating the “net” return with insurance, we see a lower expectation for ten-year returns of between 2.87% (Australia) at the high end and -0.26% (Japan) at the low end.  Needless to say, developed market sovereign returns of 1% - 3% are not going to get plans to their return targets.


Source: Bloomberg

Of course, government markets are not the only fixed income game in town.  Lower rated corporate credits, mortgage securities and the like broaden an investor’s opportunity set and we have seen dramatic evidence of plans reaching for yield over the past couple of years.  Still, like the sovereign bond markets, the current yield on these instruments, less an assumed default and recovery rate, makes for the best long-term expectations of their likely returns. 

While there are many segments within the broader fixed income universe, for the purposes of this posting, I have chosen to project returns for US High Yield, European High Yield, Emerging Market USD/EUR-Pay Sovereign and US Investment Grade segments – all fairly liquid markets.  Current Yields and Spreads (versus governments) are shown below.

US High Yield
Euro High Yield
EM Sovereign
US Inv Grade
Current Spread
Historic Spread (Median)
Current YTM
Source: BofA Merrill Lynch, Brett Gallagher Calculations

Once again, assuming no default, Current Yield to Maturity is our best guess at the long-term return from the various fixed income segments (between 4.5% and 7.0% in this example).  However, as there is some realistic level of default expected in each of these riskier pools, building a sensitivity analysis around the historic default level and recovery rates makes sense (detailed in the tables below). 

For reasonable default and recovery assumptions, I turned to data gathered by Moody’s Investors Service which has compiled cumulative 10-year default percentages across a number of markets and ratings classes for the 10-year periods ending 1970 - 2013 and also used their observed historical recovery percentages. 

I have converted the cumulative default figures into annual ones (shown in parenthesis below).  For the sensitivity analysis, I use the Worst, 25th Percentile, Median, 75th Percentile and Best Case historical experiences - though I would expect the 25th to 75th percentile range to capture the most likely outcomes.  Because of the longer history of the US data, I use that experience for other speculative markets as well (note: using a common data set with an inception date of 1983, European High Yield and EM Sovereign Debt actually have lower default rates than the US Universe over the common period – thus my assumptions may be slightly conservative).  For recovery rates, the sovereign experience has been better than corporates by about 10 percentage points and is also reflected in the sensitivities.

10-Year Cumulative (Annual Equivalent) Default Rates (1970 - 2013)

Investment Grade
US HY, Euro HY, EM Sovereign
Worst Case
10.33% (0.97%)
45.88% (3.85%)
25th Percentile
5.91% (0.58%)
38.93% (3.34%)
4.89% (0.48%)
32.33% (2.84%)
75th Percentile
3.63% (0.36%)
18.80% (1.74%)
Best Case
1.26% (0.13%)
0.37% (0.04%)
Source: Moody’s, Brett Gallagher

Using a plausible range of default and recovery assumptions, we are now able to construct a narrow range of likely outcomes for nearly any fixed income segment we desire.  While the returns due to risk assets appear relatively attractive when compared with developed sovereign markets, the range of returns are far below historic experience and most investors assumed return assumptions. 

US High Yield Debt – Yield to Maturity 5.90%, Median 10-year Cumulative Default 32.3%

Annualized Default Rates
Recovery Rate 4.31% 0.04% 1.74% 2.84% 3.34% 3.85%
30.0% 5.76% 4.74% 4.12% 3.84% 3.57%
35.0% 5.76% 4.80% 4.22% 3.96% 3.70%
37.5% 5.76% 4.83% 4.27% 4.02% 3.77%
40.0% 5.76% 4.86% 4.31% 4.08% 3.84%
42.5% 5.76% 4.89% 4.36% 4.13% 3.90%
45.0% 5.76% 4.92% 4.41% 4.19% 3.97%
50.0% 5.76% 4.98% 4.51% 4.31% 4.10%

European High Yield Debt – Current Yield 4.49%, Median 10-year Cumulative Default 32.3%
Annualized Default Rates
Recovery Rate 2.93% 0.04% 1.74% 2.84% 3.34% 3.85%
30.0% 4.42% 3.36% 2.72% 2.44% 2.15%
35.0% 4.42% 3.43% 2.83% 2.56% 2.30%
37.5% 4.42% 3.46% 2.88% 2.63% 2.37%
40.0% 4.42% 3.49% 2.93% 2.69% 2.44%
42.5% 4.42% 3.53% 2.99% 2.75% 2.52%
45.0% 4.42% 3.56% 3.04% 2.81% 2.59%
50.0% 4.42% 3.62% 3.14% 2.93% 2.73%

US Investment Grade (Baa) – Current Yield 4.65%, Median 10-year Cumulative Default 4.89%
Annualized Default Rates
Recovery Rate 3.09% 0.04% 0.19% 0.23% 0.34% 0.52%
30.0% 4.57% 4.47% 4.45% 4.38% 4.27%
35.0% 4.57% 4.48% 4.46% 4.39% 4.29%
37.5% 4.57% 4.49% 4.46% 4.40% 4.30%
40.0% 4.57% 4.49% 4.47% 4.40% 4.31%
42.5% 4.57% 4.49% 4.47% 4.41% 4.31%
45.0% 4.57% 4.50% 4.47% 4.42% 4.32%
50.0% 4.58% 4.50% 4.48% 4.43% 4.34%

 EM Sovereign (USD/EUR Pay) – Current Yield 6.84%, Median 10-year Cumulative Default 32.3%
Annualized Default Rates
Recovery Rate 5.21% 0.04% 1.74% 2.84% 3.34% 3.85%
40.0% 6.64% 5.75% 5.21% 4.98% 4.74%
45.0% 6.64% 5.81% 5.31% 5.09% 4.87%
47.5% 6.64% 5.84% 5.35% 5.14% 4.93%
50.0% 6.64% 5.87% 5.40% 5.20% 5.00%
52.5% 6.64% 5.89% 5.45% 5.25% 5.06%
55.0% 6.64% 5.92% 5.49% 5.31% 5.12%
60.0% 6.64% 5.98% 5.59% 5.42% 5.25%
In my next post, I turn to global equity markets.