Dividends aren’t a bane – they’re a boon

Should dividends be “the bane of your investment life”? Does Australia’s system of dividend imputation “force companies to behave irrationally” – that is, pay overly generous dividends, retain insufficient earnings, restrict capital expenditures and cause shareholders’ long-term returns to fall? That’s what Roger Montgomery alleges in Dividends do not make a good company – Part 1 (6 November). For good measure, in “Why the ASX 200 has gone nowhere in 16 years” (Firstlinks, 8 November), he elaborates his criticism of what he dubs Australian companies’ “dividend fetish.”

In these articles, and also in The Weekend Australian (“Why the ASX200 Has Stalled, and How to Give the Index a Kickstart,” 4-5 November), he doesn’t explicitly state but nonetheless plainly implies that the lower is a company’s and market index’s current payout ratio, the higher will be its future earnings and returns to shareholders. For this reason, he also implies that investors should eschew dividends and instead seek companies which generate high returns on equity (ROEs), retain and invest all of their earnings.

Strident claims require compelling evidence. Yet for the most part Montgomery merely asserts; he neither analyses data nor cites any research that supports his position. This article summarises my analyses of long series of valid and reliable data from Australia and the U.S. They demonstrate that his key assertions are clearly false and that his tacit recommendation, as it stands, is highly questionable and potentially damaging.

Overview

I establish five key results. First, in nominal and CPI-adjusted terms, and whether one excludes or includes dividends, the All Ordinaries Index’s rolling 15-year compound annual growth rate (CAGR) has long been slowing. Has Australia’s imputation system, and its listed equities’ seemingly resultant rising payout ratio, “caused” (as Montgomery asserts) their long-term return to sag?

My second result flatly contradicts Montgomery: since 1987, when dividend imputation was implemented, the Index’s average 15-year CAGR has exceeded its average before 1987. So does my third result: when one incorporates dividends into the analysis, the All Ords’ underperformance relative to the S&P 500 virtually disappears.

Fourthly, Montgomery’s tacit recommendation – that investors should seek high-ROE and low-payout stocks because they will outperform dividend-paying “blue-chips” – collapses under the weight of errors of logic, untenable hidden assumptions and crucial empirical falsehoods. Lastly, and most importantly, my (and others’) fifth result comprehensively disproves Montgomery’s core claim. The truth is the polar opposite of what he asserts: an increase of the current payout ratio boosts companies’ subsequent earnings, as well as their investors’ future total returns.

I demonstrate, in short, that dividends aren’t a bane of your investment life. Quite the contrary: they’re a boon and bulwark of your long-term returns. (Of course, intelligent investors already knew that!)

Let’s Start (Mostly) Charitably

It behoves me not just to acknowledge but to emphasise that, although his most fundamental claims are clearly erroneous, in his four articles (see also Dividends do not make a good company – Part 2 (9 November) Montgomery also advances three related, substantially true and collectively crucial points. 

First, “great managers are good at two things. The first is running the business, the second is allocating capital. Some might be able to run a business, but many managers are incompetent at allocating capital.” I’d add “executives and boards” to the list, and soften the point to “many executives and boards are mediocre allocators of capital.”

Secondly, Montgomery recognises “the importance of retaining profits when returns on equity (are) high, and returning them in the form of dividends and buybacks when returns on equity (are) low, declining or (have) plateaued.” I don’t disagree, but would amend this statement to: “if a company that consistently earns a high return on equity can invest retained earnings such that they generate a commensurately rate of return, then it should prune or even forego dividends. If, however, and regardless of its ROE, a company cannot consistently invest profitably, then as a rule it should retain fewer earnings and pay higher dividends.”

Thirdly, “the majority of Australian firms are mature giants with a dominant market presence.” That’s obviously untrue, but it describes most of the members of the S&P/ASX 50 Index. Of these firms, Montgomery is broadly correct: “this dominance often leaves limited room for reinvestment ...”

The problem for Montgomery, as you’ll see as I proceed, is that these three points undermine his case and strengthen mine.

Companies that retain all of their earnings, for example, will (in the absence of compelling projects and astute allocators of capital) tend to squander them on mediocre or poor investments. Such projects tend to generate commensurate results – and eventually depress rather than boost earnings and investors’ returns. Companies which retain only some of their earnings because they pay dividends, on the other hand, have fewer retained earnings to waste – and thus tend to husband them more prudently.

Next, Let’s Clarify

“Many investors and commentators,” Montgomery recently alleged in The Weekend Australian, “have been surprised by revelations of late that the S&P/ASX 200 Index has produced no capital gains in the past 16 years.”

Montgomery provides no citation for this claim, and I’ve not heard a soul mention it; nor is it a revelation to me, but never mind. Comparing the Index’s level in November 2007 and November of this year, in one respect its languor is unarguable: on the eve of the Global Financial Crisis it was 6,851; on 30 October of this year it closed at 6,773. (In another respect, however, its volatility is obvious: during this interval it fell as low as 3,338 and rose as high as 7,823 – that’s a difference of 134% from trough to peak.)

Its volatility notwithstanding, this interval’s beginning and end points’ compound annual growth rate (CAGR) was effectively 0%. The absence of long-term capital growth, reckons Montgomery (crucially, this and the other figures in this section exclude dividends), “has been a bit of a wake-up call for investors, if not an outright shock.”

He also asserts: “this isn’t the first instance of such stagnation. Australia’s stockmarket history is replete with ... multiple periods of no capital appreciation.” Actually, that’s not so; indeed, the polar opposite is closer to the truth: such periods are infrequent and comparable ones last occurred decades ago.

Using actual monthly data since January 1980, augmented by estimates from June 1937 published by the RBA, Figure 1 plots the All Ordinaries Index’s CAGRs for all rolling 15-year periods since June 1952 (i.e., June 1937-June 1952, July 1937-July 1952, ... and June 2008-June 2023). It presents these CAGRs in both nominal (that is, unadjusted for the Consumer Price Index) and “real” (CPI-adjusted, using the RBA’s data) terms.

The All Ords isn’t identical to S&P/ASX 200. Vitally, however, the former contains all of the companies in the latter, and the correlation of their monthly and annual movements is very high (r = 0.98). Most importantly for my purposes, whether directly or through valid and reliable extrapolation, the All Ords provides a much longer series than the S&P/ASX 200 (which dates from April 2000).

Figure 1: Compound Annual Growth Rate, All Ordinaries Index, Rolling 15-Year Periods, 1952-2023

Since June 1952, the Ords’ average nominal 15-year CAGR has been 6.3%; CPI-adjusted, it’s 1.2%. In nominal terms, CAGRs since mid-2022 (the lowest was 0.3% per year during the 15 years to October 2022) are the most meagre since the mid-1970s; adjusted for CPI, the most recent ones (the lowest is -2.2% per year during the 15 years to October 2022) are the worst since the mid-1980s.

At this point, comparative context becomes useful. Using monthly data compiled by Robert Shiller for his book Irrational Exuberance (Princeton University Press, 1^st ed., 2001) and updated thereafter, Figure 2 plots the S&P 500’s and the All Ords’ CPI-adjusted 15-year CAGRs since June 1952. The Australian index’s mean is 1.2%; the American index’s is 3.6%. Both series show long cycles – that is, tend strongly to regress to their long-term means – and are reasonably highly correlated (r = 0.71). Yet American 15-year CAGRs have almost always exceeded Australian ones – sometimes by big margins.

Figure 2: Compound Annual “Real” Growth Rates, All Ordinaries and S&P 500 Indexes, Rolling 15-Year Periods, 1952-2023

To see more clearly the two series’ divergences, for each rolling 15-year period since June 1952 I’ve (1) taken the S&P 500’s CAGR and (2) subtracted from it the All Ords’ CAGR. When the American one exceeds the Australian one, the resultant number is negative; when the Australian one exceeds its American counterpart, the number is positive. Figure 3 plots the results.

Figure 3: S&P 500’s CAGR Net of All Ordinaries’ CAGR, CPI-Adjusted, Rolling 15-Year Periods, 1952-2023

During the average 15-year interval, the S&P 500’s “real” CAGR has exceeded the All Ords’ by 2.3 percentage points. The series is strongly mean-regressive; both longer and shorter cycles are evident. The All Ords’ current relative underperformance relative to the S&P 500 (-6.4% per year in the 15 years to June 2023) is the most marked in 20 years and among the biggest since the 1960s.

Thus far, I’ve clarified two things that require explanation:

1. Excluding dividends and in both nominal and CPI-adjusted terms, why since the late-1980s has the All Ordinaries’ 15-year CAGR steadily declined?
2. Again excluding dividends and adjusting for CPI, why over the past few years has the All Ordinaries’ 15-year CAGR lagged the S&P 500’s by one of the largest margins on record?

Montgomery’s Explanation

“At the very least,” Montgomery wrote in his article in The Weekend Australian, “these periods of zero capital appreciation are perplexing and warrant closer examination.”

“The real revelation,” he continued, “is to be found in the reason behind the latest bout of torpor. And more shocking is that its cause is permanent, structural and likely to produce future mediocre returns ... So what gives? Why hasn’t the market done better? ... There is a protagonist so structurally strong that it overwhelms all of the positive influences (upon the market’s long-term return). In the case of the S&P/ASX 200, that protagonist is dividends.”

Montgomery outlines part of the process that underlies his assertion: “Australian businesses, especially the larger ones, exhibit a distinct trend. Propelled by the nation’s unique taxation system (namely dividend imputation), they have a propensity to distribute substantial portions of their earnings as dividends.”

Data support this contention. Before the establishment of dividend imputation in 1987, the payout ratios of the All Ordinaries and S&P 500 indexes didn’t differ significantly. Since its establishment, however, the Australian index’s payout ratio has risen gradually (from ca. 40% to approximately 70% today); the S&P 500’s ratio, in contrast, has remained stable at ca. 40% (Figure 4). To maximise the visibility of this divergence, I’ve adjusted the y-axis to exclude ratios above 100%. (During the GFC in the U.S., earnings collapsed and the ratio skyrocketed above 400%; similarly during the COVID-19 crisis and panic in Australia, it zoomed above 200%.)

Figure 4: Dividend Payout Ratios, Two Indexes, January 1974-June 2023

So far so good for Montgomery, who then adds: “dividends might sound attractive to shareholders, but they come at a cost to a company’s growth. Every dollar paid as a dividend is a dollar not reinvested in the business.” He doesn’t explicitly state them, but his implicit premises are nonetheless clear:

1. the higher this company’s retained earnings, in turn, the more it can invest;
2. the more it invests, the more it can increase its profits;
3. the more it can lift its profits, the higher will be its returns to shareholders.

In other words, Montgomery reckons that the lower is a company’s payout ratio the higher will be (notice the verb’s future tense) its eventual returns to shareholders.

As a preliminary test of this claim, I disaggregated each of the series plotted in Figure 1 into two parts: before the introduction of dividend imputation (1952-1986) and since (1987-2023). I then computed the mean CAGR for each part; Figure 5 summarises the results.

They mostly contradict Montgomery’s contention. If it were correct, then the Index’s average 15-year CAGR should be significantly lower post-imputation than it was pre-imputation. But the opposite is the case: comparing pre-1987 and post-1987, in both nominal and “real” terms the mean rises.

Figure 5: Mean CAGRs, All Ordinaries Index, Rolling 15-Year Periods, 1952-2023

Almost one-half of the 15-year intervals from 1988 to 2023 contain data from before 1988. For that reason, I’ve also computed the mean CAGRs for the 15-year periods since 2003. None of these intervals contain any data from before 1988. In nominal terms, the mean CAGR falls slightly compared to pre-imputation (4.7% versus 5.3%) but in CPI-adjusted terms it continues to rise smartly (2.0% versus -0.4%).

Montgomery is adamant: dividend imputation and the resultant rising payout ratio is a “cause” – “a protagonist so structurally strong that it overwhelms all of the positive influences (upon the market’s long-term return).” That’s rubbish. Unequivocal claims require compelling evidence, yet Figure 5 provides no such evidence. Indeed, it mostly contradicts Montgomery’s contention.

Montgomery’s First Major Set of Errors

Australia’s dividend imputation system and consequent rising payout ratios, Montgomery claims and my preliminary results contradict, explain the generally declining and now poor investment results since the GFC. What, then, to do? Montgomery implies that investors should eschew reliable payers of dividends and instead seek companies which generate high returns on equity, retain and invest all of their earnings. To make his case he offers two contrasting examples.

Examples, whether hypothetical or actual, have important uses. They remove myriad distractions and complications, and can clarify what’s usually a very messy reality.

Crucially, however, they can validly and reliably represent a population of interest only if they incorporate elements of reality whose exclusion would render the example misleading or even false. That’s the fatal flaw of Montgomery’s examples – one of which is hypothetical and one actual.

First, he describes “a hypothetical company, one consistently generating a 20% return on its equity but opting to distribute all earnings as dividends. There are no profits retained to spur growth. Consequently, the earnings and, by extension, its share price remain static over time.” This example, he adds, “isn’t just theoretical; it mirrors the behaviour of many of Australia’s so-called blue chip companies ...”

Sorry Roger, it’s simply not true that “many of Australia’s blue-chip companies” consistently generate ROE of 20% and pay all of their earnings as dividends. Indeed, depending upon one’s definition of “blue-chip” it’s hard to find more than a couple of instances of such a company. Most people would agree that Macquarie Group is a blue chip; but its payout ratio has long averaged ca. 50-60%, and never since before the GFC has its ROE exceeded 20%.

“The impact on growth and returns to shareholders,” he continues, is perhaps best explained with an example of a company that grows and pays no dividends. For that stark contrast, consider Warren Buffett’s ... conglomerate, Berkshire Hathaway, (which) has generated about 20% returns on equity for more than half a century and yet has never declared a dividend ... But the decision to retain the profits at 20% rates of return has propelled share price to a jaw-dropping $US512,000.”

Gross errors of fact and logic undermine Montgomery’s second example. According to the financial statements in its Annual Report, since 2000 Berkshire has never generated a return on equity of 20%. For the first time in 20 years its ROE approached 20% in 2019 (19%) and 2021 (18%), but plunged to -4.4% in 2022; since 2000, it’s averaged 8.1% (Figure 6). That’s a far cry from 20%!

Figure 6: Berkshire Hathaway’s Return on Equity, 2000-2022

Recall that Montgomery affirms “the importance of retaining profits when returns on equity (are) high, and returning them in the form of dividends and buybacks when returns on equity (are) low, declining or (have) plateaued.”

Berkshire’s average ROE since 2000 is exceptional by neither the standards of the average American corporation nor the typical member of the S&P 500 (see below). Moreover, its ROEs are trendless and have thus plateaued. Montgomery’s example therefore scores a spectacular own-goal: by his own standard, his poster child should actually pay dividends!

Figure 7: Return on Equity, Non-Financial Corporations, U.S., 1951-2023

In this crucial respect, Berkshire has long resembled most American corporations. Using data collated by the St Louis branch of the Federal Reserve System, Figure 7 plots American non-financial corporations’ average ROE since 1951. Importantly, this includes all corporations – not just those whose shares trade on an exchange. Since 2000, their average ROE has been 9.8%; since 2018 it’s been 8.0%; most recently (April 2023) it’s 7.6%. These means’ standard deviations are 1.5% since 2000 and 0.8% since 2018.

Accordingly, and given certain statistical assumptions, since 2018 only ca. 2.5% of American non-financial corporations have generated ROEs of more than 8.0% + (2 × 0.8%) = 9.6%

A few readers might protest: America’s listed corporations earn a higher ROE than its corporations as a whole. That’s correct: according to latest (January 2023) data compiled by the Stern School of Business at New York University, the companies that comprise the S&P 500 currently generate an average ROE of 14.5%. Equally, however, many of these entities have inflated their ROEs by purchasing their own shares – and using debt as well as retained earnings to finance these purchases.

Since 2000, according to figures compiled by Standard & Poor’s, the corporations that comprise the S&P 500 have repurchased an astounding cumulative total of more than $US9 trillion of their shares! In order to finance these purchases, they have used considerable portions of their retained earnings – and borrowed heavily.

The average S&P 500 company generates a higher ROE than the typical American corporation not least because it uses debt to retire equity. Hence it also carries a heavier load of debt as a percentage of its assets and equity. With that leverage comes risk – which Montgomery omits to mention.

Montgomery’s Second Major Set of Errors

Montgomery’s also neglects to mention that, in the absence of dividends, the prices you initially pay and subsequently receive determine your return.

Consider, as he suggests, a stock whose ROE has been 20% per year for five years. Berkshire, as we’ve seen, doesn’t qualify, and it’s vital to emphasise that few companies do. But never mind: let’s say that you locate one. Also assume (in contravention of common sense and reality, but following Montgomery) that you can safely assume that its ROE during the next 5 years will also be 20%. If you buy its stock today, what will be your return five years hence?

If you’re lucky enough to purchase it at 10 times earnings, then you’ll likely do very well: even if it still sells at 10 times earnings five years later, your CAGR will be a very healthy 15.7% per year (Table 1). Notice, though, that this CAGR is less than its ROE – hence the table’s red font. If your stock sells for (say) 20 times earnings, your CAGR rises to 32.9% – which exceeds the ROE and thus appears in green font. If the stock sells for a higher multiple, you’ll reap even more lavish returns.

Table 1: CAGRs after Five Years from Purchase of Equity Whose ROE is 20% per Year

Conversely, what if you buy the stock at 40 times and five years later it sells at just 10 times? The consequences are gut-wrenching: a CAGR of -12.3% and thus a mark-to-market loss of 1 - (1- 0.123)⁵ = 48% of your investment.

Montgomery’s enthusiasm for high-ROE and low-payout stocks prompts him to overlook the obvious: if you buy them at high multiples of earnings, then you’ll reap pedestrian or worse – relative to their ROEs – rewards (the mean CAGR is (9.5% + 3.3%) ÷ 2 = 6.4%). High-ROE stocks tend to be high-PE – and high PEs tend to fall.

On this basis, given the choice (which he offers in Dividends do not make a good company – Part 1) between a business whose ROE is 45% and a five-year term deposit whose yield is 5.1%, and given the crucial extra information that the business sells at an earnings multiple of (say) 60 times, the intelligent investor will choose the TD!

“The Arithmetic Is Irrefutable”

Yes, Roger, we heard you the first time: in response to readers’ comments to Dividends do not make a good company – Part 1, he chanted his mantra “there really is no debate; the arithmetic is irrefutable” no less than seven times!

It’s indeed indisputable: (1+ 0.20)⁵ = 2.49. In other words, a company which (a) pays no dividends and reinvests all earnings, (b) earns exactly 20% on its equity for five years and thus (c) invests each of these five year’s retained earnings at exactly the same high rate of return as its retained equity from previous years will increase its equity almost 2.5-fold, from (say) $1 to $2.49. 

Montgomery’s arithmetic is unassailable, but his grasp of logic and probability are execrable. It’s not just that the number of companies that can generate ROEs of 20% five years in succession is (as a percentage of all companies) very small; even more crucially, his – and my and anybody else’s – ability to identify these companies in advance is highly suspect. 

His examples (“suppose I send you a prospectus for an investment with a reasonable certainty of earning 45%,” etc.) simply don’t cut it. If he sent such a prospectus to me, I’d laugh and chuck it into the rubbish bin.

In other words, Montgomery’s assertion contains hidden – and untenable – assumptions. When they’re made explicit, they render his claim at best highly doubtful – and, realistically, empirically false. He’s tacitly assuming that (#1) plenty of companies meet his very demanding criterion (ROE of 20% or more per year for five years) – which, Figure 7 shows, at least in the U.S., is false; and (#2) he can identify these companies in advance – which is so doubtful that it’s effectively false.

He’s also ignoring (#3) the likelihood, if you buy their shares at a high PE in Year 1, that they’ll sell at a lower PE in Year 5; he’s also overlooking (#4) the inconvenient truth that many companies in the U.S. tend to boost their ROE by using debt to retire equity – and thus that these companies tend to be high-debt companies.

Indeed, Roger, “there really is no debate” – ignoring assumptions #3-4, and tacitly stacking effectively false assumption #2 upon false assumption #1, is patently absurd.

I happily concede that Montgomery’s abstract model (arithmetic) is undeniable. Its hidden assumptions, however, are laughable. They render his tacit recommendation (to buy companies whose high ROEs you can unerringly predict, irrespective of their debt loads, high PE ratios, etc., and on the assumption that their ROEs and PEs won’t decrease) highly questionable at best and greatly damaging at worst.

Montgomery’s Third Major Set of Errors

Hitherto I’ve excluded dividends from consideration; it’s now time to incorporate them. Figure 8, which provides a total return counterpart to the capital-only return in Figure 2, plots the All Ordinaries and S&P 500 indexes’ total – that is, capital growth plus reinvested dividends – returns. As before, it expresses these returns as CPI-adjusted CAGRs over 15-year intervals. Because the RBA’s estimates don’t contain usable measures of dividends, I’ve stuck to actual All Ordinaries data; hence the first 15-year interval is from January 1980 to January 1995.

Figure 8: Compound Annual “Real” Total Returns, All Ordinaries Accumulation and S&P 500 Indexes, Rolling 15-Year Periods, January 1995-June 2023

In two respects, “there really is no debate” – these results contradict Montgomery’s contention: first, the S&P 500 continues to outperform the All Ordinaries, but the American index’s average outperformance shrinks drastically: from 6.3% - 1.2% = 5.1 percentage points to 7.5% - 6.8% = 0.7 percentage points (Figure 9). Indeed, for a decade after the GFC, the All Ords’ total return exceeded the S&P 500’s (Figure 8).

Dividends greatly boost the All Ords’ performance relative to the S&P 500’s; as such, the high payment of dividends in Australia is a nonsensical explanation of its slightly lower average long-term total return vis-á-vis the S&P 500.

Figure 9: Mean 15-Year CPI-Adjusted CAGRs, Capital Only and Total Return, January 1995-June 2023

“With more profits retained for growth,” Montgomery wrote in Firstlinks, “it should be no surprise the U.S. market, as measured by the S&P 500, has, and will continue to, outperform the Aussie equivalent.” To see more clearly these two series’ divergences, for each rolling 15-year period since June 1952 I’ve once again (1) taken S&P 500’s total return CAGR and (2) subtracted from it the All Ords’ Accumulation Index’s CAGR. When the American one exceeds the Australian one, the resultant number is negative; when the Australian one exceeds its American counterpart, the number is positive. Figure 10, which is a total return counterpart to Figure 3, plots the results.

They also undermine Montgomery’s claim: the All Ords’ total return vis-á-vis the S&P 500 has been strongly cyclical rather than continuous. If its high payout ratio undermines its performance, then why from 2004 to 2019 did its 15-year CAGRs invariably and significantly exceed those of the S&P 500?

Figure 10: All Ordinaries’ Real Total Return Net of S&P 500’s Real Total Return, CAGRs, Rolling 15-Year Periods, 1995-2023

It’s true that since 2019 the All Ords’ total return has again – and by the greatest margin since the Dot Com Bubble of the early-2000s – lagged the S&P 500’s. Equally, however, the inclusion of dividends decreases the magnitude of its underperformance (average of 4.1% since January 2021, versus an average of 5.9% excluding dividends).

What’s causing the underperformance of the past few years? I don’t know. CAPE ratios have long been much higher there than here, but I don’t know why. I do know, however, that the All Ords’ relatively high payout ratio doesn’t explain its present or past underperformance vis-á-vis the S&P 500.

A Comprehensive Refutation of Montgomery’s Most Fundamental Contention

Montgomery’s core claim is that the higher is a company’s or market’s dividend payout ratio the lower will be (note the future tense) its subsequent earnings and returns to shareholders, and vice versa. That’s not merely demonstrably false, it’s diametrically wrong: I’ll now show that the higher is a market’s (and, by inference, a company’s) payout ratio, the HIGHER will be its earnings and thus returns to shareholders.

To test this claim, I’ve selected a favourable (from his point of view) setting: the S&P 500. As we saw (Figure 4), since the 1990s the All Ordinaries Index’s payout ratio has steadily risen and the S&P 500’s has remained roughly stable; as a result, the American index’s ratio has become ever lower relative to its Australian counterpart’s.

Over the past 30 years, in other words, the components of the S&P 500 have retained ever more of their earnings – and thus been in a position to invest ever more of these earnings – relative to the members of the All Ords.

From Montgomery’s point of view, this setting is even better: the S&P 500’s payout ratio has been trending downwards for more than 150 years; as a result, ratios since the mid-1990s are among the lowest since the 1870s (Figure 11). As I did in Figure 4, so too in Figure 11: to maximise the visibility of this trend, I’ve restricted the y-axis (in this case, excluded payout ratios above 200%. During the GFC in the U.S., earnings collapsed and the ratio skyrocketed above 400%.)

Declining payout ratios mean rising retained earnings. And if higher retained earnings beget bigger profits and better returns, as Montgomery contends, then this relationship will be apparent in the U.S. over the past ca. 150 years.

Figure 11: Dividend Payout Ratio, S&P 500 Index, January 1871-June 2023

For each month since January 1876, I computed the payout ratio’s rate of change (expressed as a simple percentage) during the preceding 5 years. I then sorted these and other relevant data by the ratio’s percentage change, divided the dataset into quintiles (that is, five subsets with equal (net of rounding) numbers of observations) and then computed various statistics within each quintile. Table 2 summarises the results. To use Montgomery’s phrase, “there really is no debate” – they show he’s diametrically wrong!

In order to make sense of Table 2, it’s important to bear in mind that the payout ratio contains a numerator (dividend per share ((DPS) and a denominator (earnings per share (EPS)). Hence the ratio is DPS ÷ EPS. Changes of the ratio over time therefore derive from two sources: (1) a change of numerator and/or (2) a change of denominator. Accordingly,

1. if DPS rises relative to EPS, say from 0.5/1 to 0.6/1, then the ratio rises;
2. if DPS falls relative to EPS, say from 0.5/1 to 0.4/1, then the ratio falls;
3. if EPS rises relative to DPS, say from 0.5/1 to 0.5/1.1, then the ratio falls;
4. if EPS falls relative to DPS, say from 0.5/1 to 0.5/0.9, then the ratio rises.

The ratio can thus fall for one of two reasons (#2 and #3) which are logically equivalent to one another; similarly, it can also rise for one of two reasons (#1 and #4) which are also logical equivalents.

Let’s turn now to the entries in the first column of Table 2, and consider a five-year interval during which the payout ratio decreased 20%, for example from 60% in Year 1 to 48% in Year 5. That interval would be grouped in Quintile #1. If the ratio increased 25%, say from 60% to 75%, that interval would appear in Quintile #5.

We can now begin to make sense of Table 2. As the payout ratio’s rate of increase accelerates (reading down the first column), several things also happen: first, the average payout ratio also rises (reading down the second column); secondly, earnings growth (expressed as a CAGR, reading down the third column) decelerates and becomes negative.

Table 2: Earnings and Total Returns, S&P 500, January 1876-June 2023, by Rate of Change of Dividend Payout Ratio

Let’s unpack this result. Quintile #1 encompasses 5-year intervals during which the payout ratio is falling. That’s because dividends are falling or earnings are rising (or both). On average, during those intervals earnings rise 11.6% per year (third column). Quintile #5, on the other hand, encompasses 5-year intervals during which the payout ratio is rising. That’s because dividends are rising or earnings are falling, or both. During those intervals, on average, earnings decrease 6.5% per year.

At this point, Montgomery might exult: “see, I told you! As dividend payout ratios rise, earnings fall!” Although that’s not his claim (crucially, he’s contending that an increase of the current payout ratio will cause future earnings and returns to decrease), such an interpretation of Table 2 is correct. As payout ratios’ rate of increase accelerates (reading down column #1) and their means increase (reading down column #2), the S&P 500’s earnings and total return decelerate. A rising rate of increase of the payout ratio and a higher mean ratio, in other words, are co-incident indicators of (1) decelerating and falling earnings (column #3) and (2) sagging total returns (column #5).

Equally, Montgomery’s core claim is patently false: over the past 150 years, an acceleration of payout ratios’ rate of increase and an increase payout ratios’ means have also been leading indicators during the next five years of (1) accelerating earnings growth (column #4) and (2) rising total returns (column #6).

What’s happening? In addition to my hunch at the beginning of this article (namely that companies flush with retained earnings will tend to squander them), here’s another interpretation: one of the best times to consider buying an index like the S&P 500 (or, by inference, one or more of its components) is when some event has smashed its earnings.

Under those conditions, (1) earnings are generally much lower than they were five years previously but (2) dividends typically fall much less than earnings; hence the payout ratio rises. During the next five years, however, earnings recover and all the while the investor receives dividends; as a result, after five years this company’s earnings multiple rebounds and its investors reap a healthy return.

Conclusions and Implications

That’s right, Roger, “there really is no debate” – dividends are absolutely vital! Of course, many investors already know from long experience that payouts underpin healthy long-term returns. My analysis provides a logical and empirical basis for their sensible stance.

My analysis is hardly iconoclastic: this basic truth possesses a long and respected – but underappreciated – pedigree. In The Theory of Investment Value (1938), John Burr Williams, a pioneer of valuation whom Buffett has lauded, wrote: “earnings are only a means to an end, and the means should not be mistaken for the end. Therefore, we must say that a stock derives its value from its dividends, not its earnings. In short, a stock is worth only what you can get out of it. Even so spoke the old farmer to his son: A cow for her milk/A hen for her eggs/And a stock, by heck/For her dividends.”

“The old man knew,” Williams added, “where milk and honey came from, but he made no mistake as to tell his son to buy a cow for her cud or bees for their buzz.” Dividends aren’t just sensible, they’re desirable: today’s high payout ratio boosts rather than trims tomorrow’s earnings and returns. This crucial fact has long been known, but today it’s not widely recognised.

My results parallel those of a landmark study. “In one of the most surprising research developments over the last four decades,” wrote Stanley Block (“The Dividend Puzzle: The Relationship Between Payout Rates and Growth,” American Association of Individual Investors, May 2009), “Robert Arnott and Clifford Asness published an article in the January/February 2003 issue of the Financial Analysts Journal entitled ‘Surprise! Higher Dividends = Higher Earnings Growth.’”

Block continued: “as virtually every textbook and article on dividends suggests that low dividend-paying stocks provide the greatest growth potential. The typical academic literature is even backed up by the ‘sustainable growth model’ measure of valuing stock prices, which suggests that future growth is largely supported by the percentage of retained earnings that is reinvested in the corporation (and not paid as dividends).”

Unlike most academics (and, apparently, Roger Montgomery), Asness and Arnott didn’t blindly worship an abstract model. Instead, they did what he doesn’t: dispassionately analyse actual data.

In their own words, “the historical evidence (for the years 1950-1991) strongly suggests that ... future earnings growth is fastest when current payout ratios are high and slowest when payout ratios are low ... Our evidence thus contradicts the views of many who believe that substantial reinvestment of retained earnings will fuel faster future earnings growth.”

“We find that low payout ratios precede low earnings growth and high payout ratios precede high earnings growth. In other words, empirically, real life acts in precisely an opposite manner to what many would forecast. Rather than future growth rising to offset a low payout policy, future growth falls when more earnings are retained” (italics in the original).

Asness’ and Arnott’s analysis, plus mine for the years 1871-2023, comprehensively refute Montgomery’s core assertion. How to explain these results? Arnott’s “favourite theory,” wrote Jonathan Clements in The Wall Street Journal (30 October 2001), ”is that cash burns a hole in management’s pocket and they over-invest in a series of less-attractive projects … The poster child for this is the massive investment in telecommunications equipment in the late 1990s. (Conversely), when they are paying out a lot as dividends, companies are forced to run lean and think very hard about every project.”

On 27 May 2001 he added: “companies are much less careful with internally generated cash than with cash borrowed from a bank or from issuing bonds. When companies borrow from outside sources, they subject themselves to the scrutiny of the capital markets. That’s a great way to discipline the way you spend money.”

Clements concluded: “more than ever, stock returns hinge on management making smart decisions with shareholders’ capital. Worried about your portfolio’s performance? You better hope the corner-office crowd doesn’t gorge itself on stock options or waste your cash on foolish empire building.”

Jeremy Siegel added a basic but crucial point. “A little more than a year ago,” at the height of the Dot Com Bubble, he wrote in The Wall Street Journal (21 August 2001), “people laughed at dividends ... (Today, however), there are increasing questions about the quality of earnings. By and large, companies that are paying dividends have to have earnings. They can’t do that with smoke and mirrors.”

“The evidence is overwhelming,” concluded William Bernstein in The Wall Street Journal (11 July 2000): “on average, a dollar of earnings reinvested in the company benefits the investor a good deal less than a dollar of earnings distributed as dividends.”

Robert Shiller concurs – and gets the last word. In Irrational Exuberance he observed:

“The reliable return attributable to dividends, not the less predictable portion arising from capital gains, is the main reason why stocks have on average been such good investments historically.”

Click LIKE so that Livewire knows that you want more of this type of content. Click FOLLOW on my profile for notification when my next wire appears.