On My Divophilia

“Classical models of finance and consumption-saving decisions predict that [a] dividend will have little effect on…consumption... Under the assumptions of Merton Miller and Franco Modigliani, for example, investors can always reinvest unwanted dividends, or sell shares to create homemade dividends, and thereby insulate their preferred consumption stream from corporate dividend policies. Thus, in traditional models, the division of stock returns into dividends and capital gains is a financial decision of the firm that has no “real” consequence for investor consumption patterns.” Baker 2007 [who then goes on to critique the proposition]  

We use the term fecundity here to refer to a portfolio's long-term ability to generate spendable cash for its owner, because fecund means "fruitful or fertile," and cash withdrawals from a portfolio are effectively its fruit… This paper will demonstrate that the fecundity of an equity portfolio — before expenses and (if applicable) taxes — lies somewhere between the earnings yield and the dividend yield of the portfolio.” Garland 2004 [1]

Dividends and Retirement – Some First Pass Thoughts


I’ve been retired for 13 years now and I have always had an unexamined – and according to others an unreasonable and un-theoried -- love for my dividend income stream (my made-up term for this is divophilia). It just feels like it is a good thing. I have rationalized myself into believing that goodness even though I am fairly conversant with financial theory and retirement finance…enough so to be quite skeptical of my own flawed inclinations and behaviors, a skepticism which might be warranted in this case. TBD

In this post I don’t think I have either confirmed or necessarily disabused myself on this topic – I have not read deeply enough nor done much wonky analytics – but I am starting to take a look at it and I wanted to think through some preliminary thoughts on paper here as a way consolidate my thoughts, thoughts which to others are probably just a “duh!” Feel free to disagree with anything here but stay respectful and constructive. I am still only half-baked on this.

So, yes, I get the theory and that there is, at least at the corp. level, a supposed indifference between dividends, buy-backs and reinvestment. Sure. Ignore for now that I have seen no small number of Fortune 500 companies destroy a crap-ton of value by retaining earnings and funneling the capital into dead-end sinkhole “projects.” But how about at the personal retiree level? Idk. Yes, we retirees can self-roll dividends by selling stock rather than getting explicitly paid. Shouldn’t make a difference they say (except for maybe how to deal with over/under spend on the div, or taxes, say). That argument is more or less classical theory. In general. Yes, I can see the point.

Otoh, keep in mind as we go into this that I have various metaphors for retirement finance. One is “driving on snow,” which I won’t touch here. The other is “goat herding” which might be more germane. Think about it. The long-term prosperity of the herding family depends on the "fecundity" of the herd in terms of fertility (reproduction, ie growth or retained earnings), production (milk, cheese, ie dividends), cull (sales of goats, butcher of goat ie self-rolled dividends), and bequeathment (handing of goats or a portfolio to the kids [no pun] at the end, ie bequest). This is the oldest portfolio management game on earth so this should all be very familiar to us. It is almost like it is in our bones to have a productive asset that is not just about implicit, retained-earnings, total-return growth. 

Goats notwithstanding, I think some of the confusion (or advocacy or fervor) on this topic might be informed by the consideration that there are perhaps at least three different intervals in portfolio management that sometimes gets elided by whomever is writing about it since we tend to focus on asset allocation so much. This may or may not help contextualize my thoughts on dividends. I'm making this up as I go but let’s at least say there is a tripartite goal-interval and objective segmentation of consumption portfolios like this: 

1. Short interval and probably hedged. Let’s say there are short-termers that will either die soon or, preferably, have entirely or mostly defeased or hedged their spend liability over a random lifetime expectation. They have a flow of life income, can spend high, and can burn the assets really, really fast, and sometimes optimally burn to zero well before death for reasons beyond this post (see LaChance 2010).  The focus here is on optimization of lifetime consumption over a front-weighted, probabilistic life. Neither div nor total returns nor even max growth are really the game afoot for this investor. Metric is probably more “discounted utility of life consumption” than not. 
   
   
2. Intermediate. Let say this is safe withdrawal rate (SWR) plans or pension schemes where life is long but not un-humanly long and where it may or may not be hedged. Here the focus is likely on medium-term investment horizons and total returns (in the context of the interval) and sometimes consumption utility optimization. Main metrics might be total returns, compound horizon growth rates, and perhaps fail rate metrics too where the magnitude of fail is evaluated in the context of a human life-scale. 
   
   
3. Long, probably unhedged. This is the domain of long dated trusts and endowments (not necessarily really perpetuities yet but maybe) -- and I think we can (should) include FIRE retirements here too -- where the focus is on a sustainable arc of spendable cash over very, very long time-frames. More on that later in the blog, but not in this post. The emphasis here is on the “utility of capital” which might be measured (according to Garland 2004 only) in terms of a portfolio's sustainable fecundity. By this he meant that the investor's capacity to draw inflation-adjusted cash is not diluted (or subjected to "shrinkage") over very long horizons that are not typical for the “average” retirement. 50y->infinity, say. (Garland: spending and "earnings plus div growth" are effectively in a time-duel; one needs to be at or under some equilibrium point) 

Given that preamble, here are some reasons why a dividend focus might sort to the top of one’s attention. This is the "pro" side: 

• Tradition/history/adages: “consume income, not principle.” Let’s keep in mind here that the “total return” idea seems like it is pretty much a child of the 20th C (a proposition I have not proven). Tell a pre-modern goat herder about total return theory. Heh. They just want milk, cheese, meat, and enough goats to pass on to the kids. Or let's consider Jane Austen in whose novels people seemed to perseverate on non-labor passive income from government bonds and real holding as a marital-and-social-status green flag. Not total-return wealth. Even Proust had some buried comments on this, too. (I mean we also have to say here that for most of human history inflation and taxes were low-ish which made everything easier.) Either way “consume income” is perhaps burned into our DNA. 
  
  
• Psych/behavioral stuff might dominate. Dividends just “feel” safer and a little bit too much like our recently and fondly (now that the safety net of human capital is mostly gone?) remembered wage income. We still often tend to think, like Austen’s protagonists, in income terms. It seems to be un-natural, for logical or human reasons, to think much in mutiplicative-infinite-horizon-consumptionless-growth-optimality terms. We could extend this to the idea of “religion” since there are very very fervent dividend advocates out there…but that is a different post. 
  
  
• Bequest goals might dominate. If one were to perseverate on leaving a bequest then low, sustainable spending of any kind – div-only being just one variety – is the game to play. Of course, no spending would enhance that even further but then we would be into institutional and/or growth optimal portfolios which is a different story. 
  
  
• Aggregate observational and cross-section analysis says mental accounting behavior does really happen. In [Baker 2007] we are mostly convinced that this (consume from div) is, in fact, how people really do behave IRL. This may be nothing more than aggregate-level confirmation of an odd behavioral bias from the first and second bullet points above, but whatever, it really happens. 
  
  
• Dividends may be a hedge on sequence of returns risk (SORR)...maybe. This is a pretty convincing argument to me and probably mostly true. If one never sells shares to fund retirement consumption, especially in a really bad down-market, then by definition we have mitigated some or all of the risk inherent in the sequence of returns if not outright created a perpetuity by spending at self-denial levels. Ignore the risks of inflation and underspending for now. 
  
  
• A focus on dividends may more than likely enhance portfolio longevity in the same way it mitigates SORR but there is an unknown effect on consumption utility in the presence of volatility and drawdowns and unnecessarily low spending. Also we have to ask if the divophilia has, ex-ante, influenced portfolio design in destructive ways. I have not studied the effect of dividends on PL but in theory (but for the "cons" listed below) one could live forever on a div-only strategy. This is probably where the fervor of dividend-only investors comes from. ERN 2020 does a great job of illuminating, though, why this might not be a truly balanced approach. I will defer to him but I am sympathetic to the fervor. 
  
  
• The Div-only approach mitigates some dilution of one’s claim on future growth/earnings were we to sell shares to fund retirement…maybe. If this is (probably) true, and I’ve seen very few make this argument (though Garland does get at this in his discussion of “shrinkage”), then the effect is likely minor outside of very high spend rates but then who knows. I’ll not dwell on this here.

Now here are some counterarguments to the div-only (DO) approach – DO for retirement income and SORR hedge – that I have run into over the years. This is the "con" side: 

• Classical theory says no, it shouldn’t matter, one can roll one’s own div by selling shares (see opening epigraph by Baker) and not only that, there may also be some efficiency and tax advantages for parties involved to skip over dividends and go straight to buybacks or earnings retention. Forget for now that for most of human history this (the total return framing) would maybe feel unnatural, 20thC financial theory notwithstanding. Also forget that at the margin, corporations pretty much suck at investing their own earnings; I’ve seen it up close and personal. (Garland: investors can’t efficiently spend 100% of retained earnings growth (EY) because some of it gets vaporized – “[billions are] slipping through investors’ fingers every year and disappearing down a sinkhole.” Oof, might be better to get paid) 
  
  
• Dividends are pretty variable, and that variability could impinge lifestyle for way too long and thus may not evaluate well in a life consumption utility context. Note that in “portfolio category/interval #1” above we might want a smoother path that can deplete wealth at either end of life or even earlier if other income is available i.e., we would spend a lot, way more than dividends, especially early in the lifecycle. ERN 2020 helpfully points out that there were several intervals in the early and mid 20th C that saw 20-50% drawdowns in div income that lasted for something like 20 years. That would a bit of a red-flag risk for 30 year retirements. 
  
  
• In a life consumption utility model, we sometimes forget that there is a utility of bequest but there are ways to add that in to evaluate the suitability or sustainability of a spend plan where bequest is either likely or planned to be >0, even for spend plans that are outside the bounds of a div-only model. A div-only plan, otoh, must confront, if not exactly the “utility,” then at least the estate planning impact of moderate to large terminal portfolios. 
  
  
• ERN (2020) makes a strong case that a DO approach is not necessarily a great inflation hedge or even inflation friendly. Yes for the last 10 years or so inflation has not been a big problem. It is now though, and it was in the 70s...I remember. Both dividend and real principal growth can be quite challenged in an inflationary environment. If we think about this carefully, we see that neither div-only nor total return portfolios hedge this risk directly. Some have made the case that a TR return portfolio is “better” in that scenario, but I have not proved that here nor is it certain. 
  
  
• A DO approach, usually because of its ex-ante portfolio design around cultivating income, often does not play -- or is a distraction to -- the growth-optimality game (not least because growth optimality is usually blind to consumption considerations). Then again, perhaps growth-optimality is not really the game played by many intermediate-horizon retirees [4]. But that is a different post. 
  
  
• Div Yields are often too low for a fun DO approach, especially now in 2021. On the other hand, ERN (2020) makes a good case that this (DY) is not a topic to be considered in isolation. In the early part of the 20th C div yields might have been higher but the growth portion from retained earnings was a little lower. Vice versa now to some degree. Check his link below to see if you agree. See 2nd bullet above, too. 
  
  
• DO myopia under-considers other sources of income and their influence on the totality of: feasibility, sustainability, consumption utility, fecundity, portfolio longevity, etc (TIPS ladders, annuities, pensions, SS etc) 
  
  
• DO may or may not actually hedge SORR all that well. I think it may in fact do so quite well but others seem to disagree. I do not have a convincing case to argue either way, but I do think that a lifestyle that is partly or fully defeased by interest and dividends will kill SORR cold. It is just that some of the other bullet points in this blog-cover might overwhelm any consideration of DO as a tool to hedge SORR. I have to think about this some more. 
  
  
• In very early retirements where the portfolio is a hybrid of many approaches and has not yet been 100% built up to where it needs to be and other income is still available, sometimes the early years div flow is just not enough now but may be ok later. ERN made this point better than I can. This point sounds weaker now that I re-read it, so idk. 
  
  
• DO is just one method among many used to goose the longevity of the portfolio (PL) and hedge SORR. Some other methods that will move the needle in the same direction – often paired with a higher consumption, btw – include stuff like: adaptive spend rules, trend following, life income, etc. 
  
  
• I have not successfully cataloged the entire topic of corp div policy here as it relates to individual retirees, but others have raised quite a few tax and transactional efficiency objections for both sides of the self-roll or div equation depending on the evaluative context. TBD. Some other post… Let’s just say that one probably has to be careful. 
  
  
• DO, in my opinion, under-considers the role that asset-based-lending can play as tool in mitigating SORR by way of adding flexibility to the timing or tranche-ing of asset sales and smoothing consumption. I mean, I seriously doubt that this would be an effective approach during 20 year drawdowns but on the margin it is, in fact, useful and I use that method fairly often. 
  
  
• DO radically under-considers an analysis about how a liquidation of the portfolio – in part or in full, now or later, irrevocable or not – and then a reinvestment or reallocation into pooled-risk-insurance (ie annuities or their close proxies), especially when evaluated in a random-life utility of consumption context, will more than likely dominate (probably by a lot) a “miserly and variable” spend from dividends-only. This alternative could be considered a form of enhanced DO if we were to frame it right and choose the right product. 

A Side Trip Down Garland’s Dividend Lane

If we were to perseverate on an un-hedged portfolio with goal/interval #3 from above (long duration consumption trust or endowment or maybe even a very early retiree) and if we were to make some weird, maybe even dysfunctional assumptions [3] then we can play a Garland (2004) game using the S&P DY and EY. I am not saying this is necessarily a great or insightful game I’m just saying we should play it for one second for fun. The game, in his terms, is to say:

a) Spending the DY is more than likely underspending or at least the minimum level that makes sense

b) Spending the earning yield or higher impinges on “fecundity” due to corporate dorkiness [2] not to mention that “Theory says that fecundity equals the earnings yield; experience says something very different.” (Garland, see also epigraph #2 at the top)

c) So, let’s spend 130% (Garland; arbitrary level) of the div yield which is supposed to helpfully be at some hypothetical fecundity-equilibrium point somewhere between the DY and EY

Applied to the S&P over the last ~60 years it’d look like this (data from multpl.com) where the orange is the EY, blue is the DY and grey dashed is 1.3 x DY. Horizontal is 3.5% as a reference:

Figure 1. DY and EY of the S&P

Then if I were to chart 1.3*DY/.035 – where .035 is just a guess at a drawdown rate that is not too small and not too big (i.e. 15% would clearly impinge fecundity over a long time frame, right?) – in order to get a sense of the Garland rule in relation to a more stable assumption, then it’d look like this. I am not sure this is helpful, I mostly just wanted to see what it looked like:

 

Figure 2. Ratio of 130% of DY : .035

Any observations from the charts? Yeah, dividend spending is volatile (but less volatile than the market values not shown) and also now look pretty darn lean in the modern era. I have not talked much about market values but certainly we have seen a lot of capital growth in the last 12 years so the crimp on div is not the whole story if one were self-rolling dividends from stock sales. But either way, Garland: “due to sharply rising stock prices, it takes much more money to buy an endowment-like spending stream today than it would have taken in 1981.” And that was from the perspective of 2004. In 1981 $19 buys a dollar of flow; it is $75 now. Ouch.


Concluding Thoughts

My divophilia is predictable, common, understandable and would be recognizable by most people for most of human history were it not for the academic finance steam-rolling of the 20th Century. I mean, yes I really do get, though, that: a) it doesn’t really matter and that total return is, in fact, the game we should probably play[5], b) I am not really an endowment and human-sized time horizons should force consumption “up” from perpetuity-like DO levels, c) div-only has been a little lean always but especially now, d) I do in fact like smoother income and expense so there is that, e) annuities would be a game changer in retirement income, but… I still like my div. I like getting paid. 

While nothing in this post is truly dispositive of anything on this DO topic – not least because I have not invested in the rigor necessary to be very dispositive – I am glad I did it because I now have down on paper some stuff that has been circling in the back of my head for a few years. Worth it just for that.



Notes

[1] I might frame this differently. I have not proven this, but it seems like the fecundity of a portfolio is likely also under the geometric mean return at some arbitrary horizon given some implementable and stable expectation for return and vol, which now that I think about, would probably be impossible to do in operating a real portfolio in real life. The idea seems about right. Ignore taxes and fees for now.

[2] I’d assert that the fecundity is hurt by spending more than the expected horizon geometric return but that is another story. Garland does not address that and barely addresses allocation to less than 100% S&P.

[3] To make this work we’d maybe have to say something like this:

- We are a retiree that never ages and/or a retiree that starts anew every year, with…

- Wealth that is unknown (but size usually does matter), and

- Other income is unknown but matters a lot, too, and we allocate to

- 100% US large cap equities (S&P) with a known div yield (DY) and earnings yield (EY), and

- We mostly just look at the annual spread between DY and EY

According to Garland these kind of assumptions “presumably overstate the fecundity of a balanced stock and bond portfolio.” Beware.

[4] A retiree with a mostly or fully hedged lifestyle can take very high risks and/or act like a long-horizon institutional investor and can play the growth optimal game to their heart’s content.

[5] Maybe…depends. Different post.




Sources, Conversations, Additional Reading, and Selected References

- Baker et al 2007, The Effect of Dividends on Consumption, Brookings.edu https://www.brookings.edu/wp-content/uploads/2007/03/2007a_bpea_baker.pdf

- Collins, P 2019, Sequence of Returns Risk Reconsidered, Investments and Wealth Institute or SSRN https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3522645

- ERN 2020, Should we preserve our capital and only consume the dividends in retirement? – SWR Series Part 40, https://earlyretirementnow.com/2020/10/14/dividends-only-swr-series-part-40/

- Garland, J 2004, The Fecundity of Endowments and Long-Duration Trusts, Economics and Portfolio Strategy, Peter L. Bernstein, Inc

- LaChance, Marie-Eve 2010, Optimal Onset and Exhaustion of Retirement Savings in a Life-Cycle Model; Cambridge U Press. 2010

- Merton Miller and Franco Modigliani, Dividend Policy, Growth, and the Valuation of Shares, 1961. The Journal of Business, 1961, vol. 34, 411

- Yaari, M 1965. Uncertain Lifetime, Life Insurance, and the Theory of the Consumer

- Private Correspondence, Francois Gadenne, CTRI-USA.org

- Private Correspondence, David Cantor, SOA

- Private Correspondence, Ken Steiner, SOA