Two key amendments to Shane Oliver’s “9 keys to successful investing”

Shane Oliver’s recent article (9 keys to successful investing - and why they are more important than ever amid COVID, 15 October) summarises “nine key things for investors to bear in mind in order to be successful.” It contains plenty of common sense (which isn’t very common these days) and even some wisdom (which is rare). It’s also concise; indeed, it’d be hard to pack more essentials into an equivalent number of words. All investors can benefit from a review of Oliver’s keys – and should adapt them to their circumstances and objectives. Arguably, however, his list isn’t complete (none, certainly including any that I might compile, could be); nor are all of his contentions indisputable (mine aren’t, either). In financial markets, only one thing is certain: nobody knows for sure! That’s why, in the spirit of respectful criticism and healthy debate, my review has inserted a crucial caveat into one of his nine keys, and added a tenth.

Revising Key #7

Oliver writes:

It often feels safe to be in a crowd and at times the investment crowd can be right. However, at extremes the crowd is invariably wrong – whether it’s at market highs like in the late 1990s tech boom or market lows like in March. The problem with crowds is that eventually everyone who wants to buy in a boom (or sell in a bust) will do so and then the only way is down (or up after the crowd panics).

The Market’s Low in March Wasn’t an Extreme

The quoted paragraph’s first and last sentences are sensible; so is most of the second one – except the assertion that the market’s low in March was an extreme. This is demonstrably false. Indeed, it’s diametrically incorrect: the low in March was the only time this year that its valuation WASN’T extreme! If my analysis is valid and my data are reliable, and Oliver’s unsubstantiated assertion is false, the implications are profound:

1.  We agree that at extremes the crowd is wrong;
2. Oliver asserts (and below I disprove) that the market’s low in March was an extreme, and I demonstrate (and Oliver seemingly denies) that today’s market is extreme;
3. Hence either (a) the crowd was wrong in March but isn’t today (Oliver) or (b) the crowd wasn’t mistaken in March but is today (Leithner).

 Figure 1 plots one measure of the All Ordinaries Index’s valuation – its price-to-earnings (PE) ratio – since January 1974. The PE’s mean is 14.2. Extremely high PEs occurred in August 1987 (21.1, shortly before the Crash of 1987), January 1994 (22.8), November 1999 (23.6, not long before the Dot Com Bust) and today (August 2020’s PE skyrocketed to an all-time high – by a country mile – of 39.8). The multiple’s most recent extreme low, which marked the nadir of the GFC and the start of a bull market that lasted until February 2020, occurred in March 2009 (9.3). This March, when the Index plumbed its low, its PE was 15.4 – which is little more than its long-term average. At this year’s March low, in other words, valuations didn’t (as they did at the nadir of the GFC) become extremely cheap. Quite the contrary: on an historical basis they remained slightly above-average.

Figure 1: PE Ratio, All Ordinaries Index, January 1974-August 2020 (Monthly Observations)

Today’s Market’s Multiple Has Skyrocketed to an Unprecedented High …

The standard deviation of the Index’s monthly PE is 3.95. Today’s is therefore an astounding 6.5 standard deviations above its long-term mean (that is, (39.8 – 14.2) ÷ 3.95 = 6.5). In plain English, and given roughly plausible statistical assumptions, only ca. three per one million events drawn randomly from a population of these events lie more than six standard deviations above or below their mean. By this criterion, the market’s current multiple has reached an unprecedented, utterly-off-the-chart extreme. 

Yet by their very nature, extremes can’t last: if one does endure, then it ceases to be exceptional and becomes a new standard! If today’s nosebleed PE isn’t a new norm, then something’s got to give: either earnings must rise dramatically (causing the multiple to fall greatly and towards its long-term average), or earnings remain depressed and the PE plummets (i.e., the market relapses spectacularly). Nobody – including me – knows whether things really are different this time; time will tell. Meanwhile, it’s essential to bear in mind that the market’s PE is no more than a crude yardstick. January 1994’s, for example, shows that an extreme multiple doesn’t necessarily foreshadow a crash (however, during the following year, although earnings surged 25% the All Ords sagged 16%); conversely, the months preceding the GFC demonstrate that a market crash doesn’t require an extreme PE.

… Partly Because Its Earnings Have Collapsed to a 20-Year Low

Today’s market, judged by a long-established criterion, has reached an extreme. Why? The All Ords’ PE has soared to an extreme high partly because its earnings have slumped to an extreme low. Most market participants – including, it seems, Oliver – are ignoring or denying the “earnings recession” (which has now become a depression) that’s long afflicted Australian equities. Figure 2 plots the All Ordinaries Index’s earnings since 1974. During the last quarter of the 20^th century, they rose steadily in nominal terms (from $35 in January 1974 to ca. $150 in December 1999) and were stable in CPI-adjusted terms (i.e., average of ca. $250 in 2020 dollars). From the turn of the 21^st century to the eve of the GFC, on the other hand, earnings zoomed: in both nominal and CPI-adjusted terms, they effectively doubled.

 Figure 2: All Ordinaries Index’s Earnings, January 1974-August 2020 (Monthly Observations)

The GFC crushed earnings: from peak (October 2008) to trough (December 2012), they plunged ca. 41% (nominally) and 46% (CPI-adjusted). For the next seven years, they fluctuated without trend.

 The Global Viral Crisis (GVC) has slammed earnings further: during the past 12 months they’ve plummeted 60%. As a result, they’ve now (August 2020) revisited the nominal level they first reached in 2001. At the beginning of this year – before the GVC – they hadn’t come close to recovering from the drubbing they’d received during the GFC. Today’s earnings ($157), which reflect the GVC’s impact, are just one-quarter as high as they were at their peak ($612 in October 2008). Note as well that the GFC’s full impact upon earnings took four years to appear. 

For these reasons, I’m deeply sceptical that earnings will soon recover from the GFC’s and GVC’s hammerings. If market participants belatedly acknowledge the prevailing “earnings depression,” then at some point the All Ords will plunge. Moreover, if something resembling an average PE reappears, it’ll collapse well below its March low.

Hence my suggested revision: “Key #7: Beware the crowd at extremes – and also that today seems to be an extreme.”

Proposed Key #10: Reinvest Dividends – and Understand the Difference between Arithmetic and Geometric Means

“The best way to build wealth,” says Oliver, “is to take advantage of the power of compound interest and have a decent exposure to growth assets.” “Although the average return on [Australian shares since 1900 is 11.6% per annum],” he continues, “is just double that on bonds (5.9% pa), the magic of compounding higher returns over long periods leads to a substantially higher balance.” I agree – but add two critical caveats.

If You Want to Compound Your Capital Then You Must Reinvest Your Dividends

Monthly estimates of the All Ordinaries Index from January 1875 to December 1979, originally compiled by Wren Research, and actual data since 1980 indicate that during 12-month periods since January 1900 the Index’s average increase is 6.8% (and since January 1875 it’s 6.2%). These data exclude dividends; Oliver’s include them. Hence my first caveat and a critical point that he overlooks (or, at least, obscures): capital gains comprise only ca. 60% of Oliver’s ultra long-term average annual total return (i.e., 6.8% ÷ 11.6% ≈ 60%); dividends provide the remaining 40%. Similarly, and as I show below, over the most recent long-term interval, the reinvestment of dividends has constituted an even higher percentage of the long-term total return.

In order to compound your capital, you must reinvest your dividends. (This doesn’t mean that you must plough the dividends you receive from X Ltd back into X; it can mean that you reinvest them into some other investment.) I’ll add a tentative corollary: if my caution about future returns is warranted, then what’s recently been true will in the future become even more so. That is, total returns won’t merely derive primarily from dividends; any return will come exclusively from them (spoiler alert: see the bottom row of Table 1 below).

Arithmetic versus Geometric Means

My second caveat concerns Oliver’s expression of returns as arithmetic means. Bluntly, this approach inflates long-term results – and thereby raises investors’ expectations unrealistically. To understand this shortcoming, consider as an example an investment whose duration is two years. Assume for simplicity that it pays no dividends. Let’s say that it gains 20% in the first year and then loses 20% in the second. What’s its rate of return? Many people would say 0%. The arithmetic mean of these two percentages, after all, is (20% - 20%) ÷ 2 = 0%. That’s not false, but it’s hardly complete. (Table 1 labels it “Gross Return.”) Examining this situation year-by-year, a different picture emerges. If you invest $100, then during the first year your investment increases 20%; as a result, at the end of the year it’s worth $120. During the second year, however, it decreases 20% such that you have $120 × (1 - 0.20) = $96. Accordingly, after two years you’ve lost an amount ($4) equivalent to 4% of your initial outlay. The average rate of return is zero, but the compound rate – as expressed by this series’ geometric mean – is negative (-2% per year).

Note as well that in order to recoup a loss, one requires a greater gain than the negative return that generated the loss. A loss of 20%, for example, requires an offsetting gain of 25%, i.e., $80 × (1.25) = $100. This confirms what investors should know but usually forget: the result you enjoy in an “up” year isn’t the most important thing: the result you suffer in a “down” year is. The truism is true: mind the downside and the upside will mind itself.

Consider now a real example. In CY18, the All Ordinaries Index fell 7.4% and in CY19 it rose 20.4%. If you invested $100 at the beginning of 2018 in a hypothetical portfolio that perfectly mimicked the Index, then (for the sake of simplicity let’s ignore dividends, brokerage, etc.) by its end your investment’s market value was $100 × (1 - 0.074) = $92.60. At the end of 2019 it was $92.60 × (1 + 0.203) = $111. That’s an average rate of return (arithmetic mean) of 11% ÷ 2 = 5.5% per year and a compound rate (geometric mean) of 5.4% per year. Clearly, these two-year results are much less exciting than the 20.4% leap in 2019!

The problem is that, goaded by “experts” and the mass media, most people obsess about short-term results expressed in terms of arithmetic means: they cheer when the market soars and fret when it sinks. Few, it seems, are able to set emotions aside and place short-term results into longer-term context. Oliver does – but he nonetheless expresses long-term returns in an unsuitable form (average rates of return, i.e., arithmetic means) rather than a more appropriate form (compound rates of return, i.e., geometric means). 

In this example, the arithmetic and geometric means barely differ. So why is their distinction important?

1. The more variable are the year-to-year results within a long-term series, the lower will be the geometric compared to the arithmetic mean.
2. Unlike its arithmetic counterpart, the geometric mean incorporates the compounding – which, Oliver rightly says, is vital over the long term – that occurs from one period to the next.
3. The geometric mean is particularly appropriate for series – such as equity markets’ returns – that exhibit negative serial correlation. In plain English, Icarus fell from the skies and Phoenix rose from the ashes; abnormally high returns today tend to beget lower ones tomorrow, and abysmal results at one point tend to rebound subsequently.

For these three reasons, the geometric mean measures long-term returns more appropriately than the arithmetic mean.

Returns Since the pre-GFC Maximum

Table 1 compares and contrasts the All Ordinaries and All Ordinaries Accumulation indexes’ returns during four crucial intervals since 2007:

1. from the pre-GFC high (March 2007) to the GFC low (March 2009);
2. from the GFC low to their all-time high (January 2020);
3. March 2007-January 2020;
4. March 2007 to the present (August 2020).

By comparing these two indexes, we can isolate the contributions of dividends and capital gains to total return; and by contrasting arithmetic and geometric means, we can surmise the variability of year-to-year results upon long-term total return, etc. 

Table 1: Investment Returns (Six Definitions during Four Periods) since March 2007

A key point emerges from the table’s penultimate and bottom rows: the Accumulation Index’s rate of return from its pre-GFC crest to its pre-GVC peak is 5.9% per year (arithmetic mean) and 4.5% per year (geometric mean); during the longest interval, from March 2007 to August 2020, the corresponding returns are 4.3% and 3.5% per year. Clearly, these geometric means are considerably lower than the arithmetic mean (11.6% per annum) that Oliver cites as an ultra-long-term average. This disparity reflects the very sharp variation of year-to year returns since the eve of the GFC: they were extremely negative during the crisis, mostly positive – sometimes strongly so – in the decade to early-2020, and very volatile this year. The All Ords’ compound rate of return from March 2007 to January 2020 (0.5% per year) is less than one-twentieth of Oliver’s arithmetic mean. The total (i.e., capital gains plus dividends) compound rate is 4.5% per year; the return from capital gains only is a mere 0.5% per year. 

Hence the reinvestment of dividends comprised (4.5% - 0.5%) ÷ 4.5% = 89% and capital gains just 11% of the total compound rate of return during these years. Owners of shares who didn’t reinvest their dividends have treaded water; net of the CPI – and especially during the ca. 13.5 years to August 2020 – they’ve lost ground. Given the volatility of year-to-year results and absent the reinvestment of dividends, these have been “lost years.” This doesn’t negate – but it does significantly moderate – the graphic which accompanies Oliver’s fifth key.

Two Key Conclusions

“Do you know the only thing that gives me pleasure?” asked John D. Rockefeller, the founder of Standard Oil who’s widely regarded as the wealthiest American of all time and the richest person in modern history (ca. $US425 billion in current dollars). “It’s to see my dividends coming in.” Whatever is the pleasure of their receipt, the benefit comes from their reinvestment. It’s through the reinvestment of dividends that investors reap the full advantage of compounding; indeed, it’s ONLY through reinvestment that the Australian market has generated ANY positive return since 2007! Reinvestment cushions losses (see Table 1’s first row); it also recoups them much more quickly. The Accumulation Index took 4.5 years (until September 2013) to recover the reverses it sustained from its pre-GFC high to GFC nadir; the All Ords took more than a decade (July 2019) to do so – and since then this recovery has disappeared and long-term losses (excluding dividends over a 13-year period) have returned.

Warren Buffett first told us – and Shane Oliver has reminded us – that investors should “be fearful when others are greedy and greedy when others are fearful.” Investors (“speculators” is probably a more apt term) are now greedy. Or, at any rate (and judged by the market’s extreme multiple and earnings depression), few are fearful. 

That’s why investors worthy of the name should defy today’s crowd. If they don’t, their compound rate of total return (dividends reinvested) during the next decade or more risks falling well below 3.5% per year.

In the current extreme environment, and to adapt Oliver’s words, perhaps the avoidance of loss will be “the key thing for investors to bear in mind in order to be successful.”