Shane Oliver believes that during the year to come Australia will likely avoid a recession (see, for example, Seven reasons why Australia should avoid a recession, 9 November). But it’s quite possible that it won’t: on 28 October in Shane Oliver’s guide to riding out a “year to forget” he reckoned that “the chance of recession in Australia is 40%.”
Owners of Australian shares mustn’t let this possibility – whose magnitude nobody knows – distract them. Whether this country does or doesn’t avert a recession is largely beside the point: what affects their results much more is what happens globally and particularly in the U.S. – and how investors navigate the rapids.
In my last article (Recessions usually crush shares – but investors can always reduce their ravages, 31 October), I demonstrated that (1) investors can’t accurately predict the timing (still less the duration or severity) of recessions; (2) economic downturns in the U.S. crush stocks – not just in that country, but also in Australia; and (3) a diversified portfolio of stocks and bonds can greatly mitigate recessions’ harmful effects.
In this article, I elaborate these conclusions. I reconfirm the well-known (but inconvenient to and thus mostly disregarded by the mainstream) result that American stocks, as measured by the S&P 500 Index’s Cyclically-Adjusted PE Ratio (CAPE), have generally been very expensive since the mid-1990s. Indeed, in 2021 CAPE was higher and thus stocks were dearer than at any time over the past 150 years except the Dot Com Bubble of the late-1990s and early-2000s. To this result I add three new or unfamiliar ones:
- Australian equities as measured by CAPE have also been relatively expensive since the mid-1990s – and during 2021 were almost as dear as they were just before the Crash of 1987, the bursting of the Dot Com Bubble and the eruption of the GFC;
- American and Australian stocks typically suffer punishing losses when recession and overvaluation occur simultaneously;
- whether or not a recession occurs and given their high current CAPEs, the prospective returns of equities in Australia and the U.S. over the next ten years are – relative to their recent and long-term average returns – mediocre.
In short, the downside risks to equities in Australia are presently significant – and certainly much greater than the consensus admits. Indeed, so-called experts apparently know neither the Australian CAPE’s current level nor its historical average or evolution over the years. That’s because they haven’t bothered to compute them (or if so, are keeping the results to themselves); accordingly, short-term and long-term risks to stocks are much greater than the mainstream understands.
How low could Australian and U.S. stocks go in 2023? I demonstrate that if a recession occurs and past is prologue, investors can expect the All Ordinaries to fall up to 20% and the S&P 500 up to 30% from their respective closes on 11 November.
Yet there’s also good news: First, in the short term a conservative and diversified portfolio of stocks and bonds greatly mitigates the harmful effects of both recession and overvaluation. Second, at times like the present when equities are generally overvalued, and whether or not a recession is nigh, this portfolio offers the credible prospect of long-term outperformance compared to an all-shares portfolio.
My Most Important Point
It’s vital to emphasise at the outset that my purpose is NOT to stoke anxieties; still less is it to induce anybody to panic, liquidate his portfolio and run for the hills. Instead, my objective is to encourage you to think rationally and act sensibly. To do so, you must think and act for yourself. Unfortunately, most people automatically accept the consensus, and very few are prepared to consider reasoned alternatives – and, when necessary, ignore and even defy the mainstream. Above all, my aim is to illustrate Benjamin Graham’s maxim: “you are neither right nor wrong because the crowd disagrees with you,” he said. “You are right because your data and reasoning are right.”
“What’s required,” Warren Buffett added in Berkshire Hathaway’s Annual Report (1990), “is thinking rather than polling. Unfortunately, Bertrand Russell’s observation about life in general applies with unusual force in the financial world: ‘Most men would rather die than think. Many do.’”
The Cyclically-Adjusted Price-to-Earnings (CAPE) Ratio
The cyclically adjusted price-to-earnings ratio, commonly known as CAPE or the Shiller PE, measures the valuation of stock markets. As a measure of current valuation and estimator of prospective returns, it’s far superior to the ubiquitous “forward” PE ratio (for details, see How experts’ earnings forecasts harm investors). Devised by an American economist, Robert Shiller (see my previous article for details), CAPE’s numerator (“price”) is the level of an index such as the Standard & Poor’s 500 at the end of a given month; and its denominator (“earnings”) is the average over the preceding ten years of the index’s earnings, adjusted for the Consumer Price Index, during that month.
Benjamin Graham’s and David Dodd’s classic text, Security Analysis (1934), originated the idea of using the CPI-adjusted average of earnings over the last decade: it tamps the impact of business cycles and other events (such as a single’s year’s exceptionally good or bad results, etc.), and thereby gives a truer picture of earnings and their trend.
The higher above its long-term average CAPE rises, the more expensive (and, arguably, overvalued) is the market; and the further below its average it falls, the cheaper (and, by inference, undervalued) it becomes. Moreover, research by Shiller and a colleague, John Campbell, first demonstrated that over time CAPE regresses towards its long-term mean: at any given point in time, the higher it rises above (or falls below) its long-term mean, the greater is the likelihood that it subsequently regresses towards and below it, and vice versa.
Accordingly, CAPE’s current value, plus the assumption that it will continue to mean-revert, permits us to estimate equities’ future long-term returns: a higher-than-average CAPE today implies below-average returns over the next 10 years, and vice versa.
Although it wasn’t devised as a predictor of short-term returns (and still less a warning signal of impending bear markets, crashes, crises and panics), very high CAPEs have preceded such events. In “The Mystery of Lofty Stock Market Elevations” (The New York Times, 14 August 2014), Shiller observed that the CAPE prevailing at that time (26) had risen to “a level that has been surpassed since 1881 in only three previous periods: the years clustered around 1929, 1999 and 2007. Major market drops followed those peaks.”
American and Australian Equities’ CAPEs
Using data compiled by Shiller, Figure 1 plots the CAPE of the S&P 500 Index. Since January 1881 it has averaged 17.3 and since January 1995 it’s averaged 27.8.
Measured by CAPE’s mean since 1881, not since the depths of the GFC has the S&P 500 been moderately cheap or even reasonably priced. Measured by its mean since 1995, during the Global Viral Crisis, stocks briefly became moderately priced; presently (CAPE of 28.2 in September 2022) they aren’t unduly expensive.
Why has the S&P 500’s (and, as we’ll see, the All Ordinaries Index’s) CAPE been much higher since the mid-1990s? This period corresponds, albeit roughly, to the “Greenspan put.” According to Investopedia, “the Greenspan put was the moniker given to the policies implemented by Alan Greenspan during his tenure as Federal Reserve Chair (and continued by his successors). The Greenspan-led Fed was extremely proactive in halting excessive stock market declines, acting as a form of insurance against losses, similar to a regular put option.”
Figure 1:Cyclically-Adjusted PE Ratio (CAPE), S&P 500 Index, Monthly, January 1881-September 2022
Can central banks indefinitely inflate stocks? In particular, does CAPE’s much higher mean since the mid-1990s imply (to borrow the infamous phrase that an economist at Yale University, Irving Fisher, used in September 1929) that American stocks’ valuation has “reached what looks like a permanently high plateau”? If so, has CAPE ceased to revert to its mean since 1881? As in the late-1920s, so too during the Dot Com Bubble of the late-1990s and early-2000s: many people were utterly certain that the internet had spawned a New Era – as well as a New Economy freed from supposedly outmoded norms, and stock prices to match.
Also at that time, influential economists proclaimed that their policies had conquered the business cycle, and therefore that recessions had become things of the past. Ironically, the so-called Information Age generated huge quantities of nonsense (including absurdly unrealistic expectations), relatively little knowledge and practically no wisdom. Then the bubble popped, the GFC erupted and CAPE collapsed (albeit not for long) below its long-term mean.
After the GFC, CAPE again skyrocketed – and some experts once more asserted, in effect, that (for various reasons, including central banks’ policy of suppressing interest rates to their lowest levels on record), it couldn’t and therefore would never again regress to its long-term mean. Not least because hyper-interventionist policies’ invidious consequences have become abundantly clear – and CAPE has recently sagged – I strongly doubt their assertions.
In January 2021, the S&P 500’s CAPE exceeded 35; by the end of the year, it rose as high as 39. Those were among the highest CAPEs – and markets last year ranked among the most egregious periods of overvaluation – on record. Last year, the S&P 500 was significantly more overvalued than it was on the eve of the GFC. Indeed, in only 1.9% of the months since January 1881 (all of which occurred during the Dot Com Bubble) did CAPE exceed last year’s level. This year, it’s fallen significantly; presently (September 2022) it’s 28.
In sharp contrast, stocks in the U.S. were extremely cheap during the Depression of the early-1920s (see also Why we need a “good” depression), the depths of the Great Depression and the late-1970s and early-1980s.
If CAPE’s mean from 1881 denotes fair value, then a CAPE of 17.3 today would necessitate a plunge of the S&P 500 of more than 30% from its current level. If CAPE’s mean since 1995 represents fair value, then it must drop 15%.
Figure 2: Cyclically-Adjusted PE Ratio (CAPE), All Ordinaries Index and Predecessors, Monthly, June 1947-September 2022
Using data compiled by the RBA (see my previous article for details) for the years 1937-1975 (which I’ve merged to data since 1975 that I’ve compiled), Figure 2 plots the CAPE of the All Ordinaries Index and its predecessors. (In order to maximise the comparability of the RBA’s quarterly data to Shiller’s monthly data, I’ve converted RBA’s quarterly data into monthly observations through linear interpolation.) Since January 1947 the Australian CAPE has averaged 14.5, and since January 1995 it’s averaged 17.9.
As in the U.S., so too in Australia: stocks tend to plunge not long after CAPE rises to extreme levels. The Australian CAPE scaled its all-time high on the eve of the GFC, and its second-highest levels just before the Crash of 1987 and the Dot Com Bubble burst. It also soared during the mining boom of the late 1960s. And also like the U.S., the Australian CAPE was very low – and thus stocks were very cheap – during the late-1970s and early-1980s.
During and for several years after the GFC, Australian stocks were reasonably priced and sometimes even moderately cheap – and during the Global Viral Crisis they briefly touched fair value. Yet CAPE soared well above both of its means after the GVC. In 2021 it averaged 20.8 (which ranked its months among the top 11% since 1947); this year it’s averaged 19.8 (14.3%) and in September it sank to 17.7 (23.7%). Presently it isn’t as elevated as it was before the GFC, Dot Com collapse and Crash of 1987 – but stocks’ degree of overvaluation is comparable to the late-1960s – which preceded the collapse of the 1970s.
If the Australian CAPE’s mean from 1947 denotes fair value, then a CAPE of 14.5 today would necessitate a decrease of the All Ordinaries Index of approximately 5,850 – which implies a drop of 20% from its close on 11 November. If CAPE’s mean since 1995 represents fair value, then the All Ords must drop to ca. 6,750, i.e., 8% from its current level.
Figure 3 plots the American and Australian CAPEs since 1947. The two series are strongly correlated (r = 0.74). Until the late-1980s (the eve of the Crash of 1987 was the only time that the Australian CAPE significantly exceeded the American one), they tracked one another very closely. During the remainder of the 20th century, however, the Dot Com Bubble inflated the S&P 500 far more than the All Ords; when the bubble burst, however, the American CAPE crashed – and once again coincided with its Australian counterpart. The Australian CAPE again tracked the American one before and during the GFC. Since then, the American one has greatly outpaced the Australian one. If Australian stocks are moderately overvalued, then American ones are greatly so.
Figure 3: CAPE Ratios, Australia and the U.S., Monthly, June 1947-September 2022
Is Recession Nigh?
In The Wall Street Journal on 30 October (“Why a 2023 Recession May Be Mild”), Alan Blinder, a professor of economics at Princeton University and former vice chairman of the U.S. Federal Reserve (1994-1996), wrote some very candid and wise words: “I am often asked these days whether I think the Federal Reserve will push the economy into a recession and, if so, whether that recession will be mild or severe. The only honest answer is ‘I don’t know.’ Neither does anyone else. Economists have never been good at predicting recessions …”
But the inability to prophesy hardly dissuades experts! Over the past few months, the developed world’s central banks, major financial institutions and governments have hinted or warned overtly that the risk of recession is significant and rising. Moreover, a growing chorus of commentators and economists are predicting that a recession – perhaps long and severe one – will commence within 12 months.
On 8 November, for example, in prepared remarks to the Senate in Canberra, the Treasury Secretary, Dr Steven Kennedy, warned of a “marked” deterioration of the global economic outlook. Indeed, it’s “becoming probable that major developed economies (such as Britain and the EU, the U.S. and Japan) will soon experience recessions.”
Blinder concluded: “when pressed – which also happens frequently – I’m in the mild recession camp. The Fed would have to be both incredibly skilful and lucky to avoid any recession at all; everything would have to break its way. But there are several reasons to think that the 2023 recession, if it comes, may be mild.”
The Consequences of Recession and Valuation upon Short- and Long-Term Returns
Let’s assume that a recession occurs in major northern hemisphere economies, but not necessarily in Australia, next year. What would be the consequences for stocks – and particularly Australian equities? The analysis I summarise below demonstrates that two factors crucially impact investors’ short- and long-term returns. The first is the market’s valuation (CAPE) at the starting point of any measurement of returns; the second is whether that point is or isn’t in a recession as defined by the National Bureau of Economic Research (NBER; see my previous article for details). In brief,
- the higher is the CAPE at any given point in time, the lower are subsequent short-term and long term-returns;
- if a given month isn’t in recession, the higher are previous and subsequent returns;
- if a given month is in recession, the lower are previous and subsequent returns.
Crucially, the harmful impacts upon returns of high valuation and recession reinforce one another. As a result, if a given month is (a) in the highest quintile of CAPEs and (b) in recession, then subsequent returns are strongly negative in the short term and virtually zero in the long term.
I sorted relevant data for each month since January 1881 in Shiller’s dataset (that is, each month for which a CAPE can be computed) by CAPE. For each month, I also calculated the S&P 500 Index’s return (capital only, i.e., ignoring dividends) during the previous 12 months and the next 12 months. I then divided the dataset into five equal (by number of observations) segments. Quintile #1 contains the 20% of observations with the lowest CAPEs, Quintile #2 the next 20%, … and Quintile #5 the 20% of observations with the highest CAPEs. Finally, I computed each quintile’s average returns for both the previous and subsequent 12 months. The first four columns of Table 1 summarise the results. I also did the same for the Australian data; the first four columns of Table 2 summarise the results.
Table 1: Returns of the S&P 500 Index Stratified by CAPE, Rolling 12-Month Periods, January 1881-September 2022
Valuation underpins returns: the higher is the CAPE, the bigger are the previous returns – and the smaller are the subsequent ones. If I knew nothing except the S&P 500’s current CAPE (which lies within Quintile #5), I’d expect it to rise modestly (5.0%) during the next 12 months. Similarly, if I knew nothing except the All Ordinaries’ current CAPE (which lies towards the upper bound of Quintile #4), I’d expect an even more modest (1.9%) return during the next 12 months.
Table 2: Returns of the All Ordinaries Index Stratified by CAPE, Rolling 12-Month Periods, June 1947-September 2022
But we can bring additional information to bear. Investors can’t reliably anticipate recessions, and neither I nor anybody else knows whether the U.S. will enter into recession during the next year. Still less does anybody know the duration or severity of any such recession. But I know the impact of past recessions upon equities’ returns – and can therefore gauge the possible impact of a recession in 2023, if it eventuates.
To each of the observations in Shiller’s dataset, I merged NBER data (for details, see my previous article) that distinguish months in recession from those outside recession. I then stratified the data into quintiles according to CAPE, and sorted each quintile into “recession” and “non-recession” observations. I then computed average returns for the previous and subsequent 12 months; the seventh and eighth columns of Table 1 and Table 2 summarise the results. Regardless of the quintile of CAPE, during the 12 months before months in recession, the S&P 500 suffers losses. The same is true for Quintiles 1-3 in Australia.
In both countries, valuation matters: for recessionary months with very low CAPEs (Quintile #1), average return during the subsequent 12 months is excellent (18% in the U.S. and 12% in Australia); but in recessionary months with very high CAPEs (Quintile #5), the average return during the subsequent 12 months is terrible (losses of ca. 20% in both countries).
In other words, if a recession occurs when stocks are cheap, or if it quickly smashes them to cheap levels, then the average return during the next 12 months is very good; but if a recession occurs when stocks are expensive, or if the recession fails to reduce them to a cheap level, then on average the next 12 months’ return is awful.
If past is prologue, and given the two indexes’ current CAPEs, if during the next year the U.S. falls into recession, then during the subsequent 12 months we can expect stocks in Australia to sag ca. 12% and the U.S. approximately 20%.
What if there’s no recession? Given today’s high CAPEs in both countries, expectations regarding returns should be modest. Specifically, Tables 1 and 2 tell us to expect single-digit returns ranging from 3-4% in Australia to 7.4% in the U.S.
I replicated the foregoing analysis for rolling periods of 120 months, and expressed returns as compound annual growth rates (CAGRs). Table 3 and Table 4 summarises the results. Their first four columns confirm that returns don’t merely regress over 12-month periods; they also regress over ten-year intervals.
Valuation remains crucial to returns: the higher is a given month’s CAPE, the bigger was the CAGR over the preceding ten years – and the smaller will be the CAGR over the next ten years. This result applies in Australia and the U.S., and whether the month in question is or isn’t in recession.
Table 3: CAGRs of the S&P 500 Index Stratified by CAPE, Rolling 120-Month Periods, January 1881-September 2022