Jump to content

What is your prediction for range for peak Shiller PE this time?

Recommended Posts

It is hard to predict when the market will crash next. 

I wonder if it will be easier to predict the range for peak Shiller PE this time? 

I think we can probably all agree with almost 100% probability that we won't hit Shiller PE of 100.  We can probably also agree that probability is high that we won't hit Shiller PE of 50.  What about 40 and 45?  Above what number will folks start getting some more dry powder ready? 

Having a range in mind for max Shiller PE might help us be prepared. 

So far we have went above Shiller PE of 40 less than 2 years in 150 years, i.e. less than 1.3%.  Only the 2000 crash started at Shiller PE above 40.  All other crashes were at Shiller PE of 30 or below, except the 2020 flash crash.  I understand we are in different times with respect to interest rates - if you think these times are different because interest rates are low, it will be great if you could consider that over the past 150 years, we haven't had interest rates low for too long either (https://www.multpl.com/10-year-treasury-rate). 


Source: https://www.multpl.com/shiller-pe

Edited by LearningMachine
Link to comment
Share on other sites

  • LearningMachine changed the title to What is your prediction for range for peak Shiller PE this time?

My issue is this doesn't adjust for the fact that corporate tax rates were 35% for 7 of the last 10 years and then 21% for 3 years (of which 2020 was one). If I normalize for 21% tax rate CAPE is closer to 31% which is still quite high but interest rates are going to be low till 2023. I am in 20% cash but will sell more if S&P crosses 4500 this year. 

Link to comment
Share on other sites

I don’t really see a big correction right away. As Jamie Dimon says, we are entering a Goldilocks period of multiple tailwinds for the economy. There is more risk in staying out than staying in here I think. 

As for interest rates, I don’t see them much higher than where they are. The perennial QE and low interest rates have taken OECD debt to GDP to such a level that higher interest rates are just not an option anymore (so long as the central banks can help it). I’m sure there will be market sneezes along the way but nothing crazy. We are swimming in a sea of massive liquidity and the pro cyclical forces continue to gain ground.

Edited by Ice77
Link to comment
Share on other sites

  • 2 months later...
Posted (edited)
On 4/14/2021 at 9:57 AM, adesigar said:

My issue is this doesn't adjust for the fact that corporate tax rates were 35% for 7 of the last 10 years and then 21% for 3 years (of which 2020 was one). If I normalize for 21% tax rate CAPE is closer to 31% which is still quite high but interest rates are going to be low till 2023. I am in 20% cash but will sell more if S&P crosses 4500 this year. 


This is a very good point. But if corporate rate goes up to 28% AND inflation/interest rates increase, that's a double whammy to the market that could start a slow avalanche.


The way I look at it is given near zero interest rates and record low tax rates the Schiller PE should something like 50% higher than normal, but instead it's more than double.  I believe that happens because as prices gradually rise to their correct valuations for lower interest and tax rates, they tend to overshoot. Everyone is making money in the market so money piles in and people take lower yields because thats whats winning (and FOMO from those late to the game), and thats how we get to a Schiller 38.


But when those valuation factors reverse and people start dealing with losing months and a losing year, it works in reverse. Even if interest rates only go up to 3% and taxes to 28% and a "fair" average PE is still around 20 or so once the market hits that level it might keep going down for a while as no one wants anything to do with stocks after a 50% loss.

Edited by ValueArb
Link to comment
Share on other sites

One other thing that seems to have happened during the pandemic. A lot of small businesses were hurt really bad but it seems the large companies are mostly doing much better. Some earnings seem to have moved to large companies. Recent revisions to S&P earnings estimates are $180-$190 for 2021 and $200-$225 for 2022. That’s a forward PE of 20.

Link to comment
Share on other sites

I don’t think rising rates (if they occur) will directly and materially affect earnings for a very long time. The majority of large cap publicly America  investment grade and is 


a) underlevered


b) is borrowing on an average of 17 yr maturity at issuance (SIFMA)


c) has debt that has an average maturity of 13.75 years (LQD etf = source)


rising rates maybe create an alternative to equities but not for a a few hundred bps (sooner for liability matchers like frozen pensions).


so I see no margin mean reversion from rates.


Going from 20% to 30% tax rate would be a 12.5% decreases in earnings power. ($80 vs $70 of post tax earnings for every $100). $10/$80 = 12.5%. if this came to pass and the whole market went down 13%, I wouldn’t regret being 100% long. Likewise if we land in between and earnings go down 7%. 

so in my view rates and taxes are boogeymen meant to separate bears from their money over the long term. 


i worry more about other stuff, but ultimately find most stocks to not be egregiously valued. I’ve thought corporate America is over-earning for a while, but one has to believe it’s REALLY overearning to be REALLY concerned with valuations.


so I don’t know if I really care at all about Shiller PE and it’s mean reversion based metric.

Why is MSFT or AMZN or FB’s average of the last 10 years earnings relevant? Entire new business lines have been created in the past decade. 

Link to comment
Share on other sites

The Shiller isn't a P/E for an individual company, its for the whole market. And while big tech has grown earnings like crazy, entire industries and all their earnings have vaporized in that time (coal, department stores/lots of retail, shale gas, on premises software, etc), which is at least partially offsetting.


I dont disagree with your general assessment, even though 2020 S&P earnings were lower than 2011 (obvious covid effects). But I do think the market is pricing in continued strong growth from 2019 levels.


It doesn't seem obvious that will happen to me. Labour taking a bigger share of the pie seems like a particularly significant risk factor.

Link to comment
Share on other sites

If I froze up every time in my investing life that there was some perceived catastrophic risk I'd be nowhere. Theres folks out there who have called 29 of the last 2 bear markets. More money has been lost preparing for bear markets than has been lost in bear markets. I always try to remember these things because getting too caught up in the noise and worrying about every little thing and then placing outsized(and outrageous) probabilities on every risk is detrimental to ones wealth. 

Link to comment
Share on other sites

How could anyone be 100% certain that CAPE wouldn't hit 100? If rates go negative, I don't see why that couldn't happen. Not saying that it will but the odds are much higher than 0%. You factor in negative rates (and no one no knows how that will impact other things), you factor in government support, human emotions, etc, hitting 100 isn't totally out of the question (it's unlikely but certainly possible).

Link to comment
Share on other sites

2 hours ago, stahleyp said:

How could anyone be 100% certain that CAPE wouldn't hit 100? If rates go negative, I don't see why that couldn't happen. Not saying that it will but the odds are much higher than 0%. You factor in negative rates (and no one no knows how that will impact other things), you factor in government support, human emotions, etc, hitting 100 isn't totally out of the question (it's unlikely but certainly possible).


Europe and Japan have had 0 to negative rates on and off for years now. No 100 CAPE ratios there. 


At sufficiently low enough rates, the theory and application behind discount rates just completely falls apart. 

Link to comment
Share on other sites

  • 3 weeks later...

It’s risky to enter a discussion where the title contains « prediction » and « peak » and where a humble attempt at fundamental analysis may be interpreted as an attempt to spoil the fun.

The idea here revolves around the link between earnings and the multiple that should be attributed to such earnings based on expectations of both components and to see if there is any relevance with the CAPE topic.


And it’s particularly tricky when the line of thinking challenges previous long term trends that appear to aim for infinity, and beyond.


Short version:

NIPA profits show core corporate profits and take into account the full cost of stock-based compensation which GAAP does not, suggesting that present-day ‘traditional’ value measures such as PE and CAPE ratios based on reported earnings are, in fact, materially higher than officially reported.



Optional section and longer version:

The sub-topic of interest here, for the market as a whole and also for individual stock picking, is the ‘quality’ of those reported earnings in the sense of their connection to true core earning power and more to the point the apparent disconnect that has built up between NIPA profits and the stock market GAAP reported earnings.

This topic is neither exactly mainstream nor exciting, at least on the surface. Also, when occasionally mentioned by some, the subject tends to be accompanied by superficial analysis and the disconnect tends to be attributed to the notion that there are now a few dominant growth companies that benefit from intangibles, brands and economies of scale that skew profitability results and to the associated notion that there should be an appropriate multiple applied to such great prospects. This is a kind of topic where digging into numbers reveals very different conclusions.

In the late 1990s, i looked into Microsoft and Intel and some related others and figured out they were out of my league (the main reason related to an inability to properly understand the business but there were issues with stock-based compensation). That’s too bad because, eventually, Microsoft’s stock price chart became parabolic. That’s OK and I’m happy for those truly enjoying the more recent ride. In those early days, when options were not really expensed, especially during the vesting period, i learned a few nuggets of investment knowledge going through the mechanics of stock compensation accounting. This was a time when, in some companies especially of the dot-com flavor, a large percentage of free cash flows was allocated to buyback activity and through detailed dissection of footnotes and retrospective feeding into missing financial elements, i was able to figure out that the eventual cost of stock options given to management ended up, at least for a while and especially during the parabolic phase, way higher than what GAAP allowed or forced to reveal. i recently learned that the NIPA-GAAP reporting disconnect was also a correlated element then and there may be relevance for today. Over the years, i’ve often done this type of work on specific issues when stock-based compensation and buyback activity were significant components and the disconnect between the periodic amortized cost of the reported “fair value” of this compensation can be way less than the ultimate true intrinsic value of the harvest. Of course, there are times when the ultimate value is ‘deserved’ (IMO not often, small characters).


This post has likely reached a stage where absolutely nobody is reading but, to make sure, here’s a short review of the potential significance of this NIPA disconnect and the potential link to present-day CAPE.


NIPA corporate profits correspond to profits reported by all corporations (public and private; large and small). The numbers are reported using a basic method which includes some estimation and sampling and then, in the following years, numbers are adjusted when data crystallizes towards the true numbers (at least to adequately reflect what NIPA is supposed to measure). It’s the type of stuff that eventually feeds into GDP numbers etc. Compared to GAAP reporting (S&P 500, Russell 3000 or whatever), the NIPA corporate universe is different because it includes also private and smaller businesses. Also, the data collection methods as well as the measuring methods are different. For example, during downturns, GAAP reporting will tend to under shoot and then to go back to the NIPA trend during the recovery but, over time, the long-term trends are (and should be) broadly similar, especially when comparing S&P 500 operating earnings and NIPA national profits after tax (with some typical adjustments: “without IVA and CCAdj”). That very tight correlation manifested itself up until the late 1990s. Then there have been two potentially unexplained periods (the topic of this post) of significant disconnect, one around the dot-com bu**le and another since around 2012. Why (and the potential utility for the whole market assessment or of specific issues) is the topic at hand.

GAAP has specific rules and one could argue that there is a growing amount of management discretion and that’s something to consider. Also, it is being increasingly suggested that GAAP has become a poor tool to measure and report the growing importance of ‘intangibles’ but that’s another story.

NIPA profits reported over time, similar to a depreciation concept and tax accounting, are characterized by temporary differences that get resolved over time, such as, for example, when they eventually factor in the tax assessment results that are eventually ‘discovered’ (in the sense of price discovery) in order to actualize results of previous years. They typically report revisions on prior years’ results during the summer (this summer should be very interesting) and sometimes revisions are quite large. In general, in the last few years, revisions have resulted in a significant decrease in reported corporate profits (lower ++ in 2019, higher + in 2020). So, the last reported years up to two to three years are less reliable. NIPA profits can also be characterized by permanent differences which is potentially interesting because it likely primarily reflects the permanent difference resulting from the NIPA recognition of stock-based compensation that is never fully recognized by GAAP.

There are many lesser issues (pension adjustments etc) which may produce permanent differences but those have not been historically associated with material differences. Another source of permanent difference are large write-downs that tend to get reported in GAAP (not NIPA) during major economic downturns. This was especially significant during the GFC but this aspect is not relevant for the 2011-9 period or not likely relevant for the more recent sharp down-up viral-related events.

The major source of permanent difference is related to stock-based compensation.

A foray into stock compensation accounting since they have been expensed reveals that GAAP accounting for about the last 15 years requires to estimate the fair value of the stock options allocated upon grant and to allocate this value over the vesting time period, irrespective of the ultimate value realized. Similar principles apply for stock awards (fair value calculated at grant date and then expensed over vesting or service period). The ultimate value could be less than initially estimated for example because more people leave the company than planned giving rise to an unusual level of options forfeited. This is usually accounted for, from the start and adjusted afterwards. The ultimate value could be less or more, depending on the actual evolution of the underlying stock price (and on the timing of the owner’s sale of the stock) and this is where NIPA profits come in handy. Because the NIPA corporate profits method eventually integrates tax data and the value eventually realized by holders through tax data, the ultimate value realized is captured and this helps to determine who benefits from this value and who pays for it. Lately (since about 2011-2012, ie after the noise related to the post 2007-9 recovery), the GAAP profits have diverged from NIPA profits and although there may be several explanations, most likely and by far, i think this means that reported GAAP earnings have overstated core underlying earnings as stock prices have done unusually well and as NIPA has been showing the eventual value realized from stock options (and stock awards) and how this should be deducted as an additional items in operating expenses of prior years. Note that the full realization of this disconnect takes time.

Here’s the standard reported NIPA profits chart on which is overlaid the equivalent pre-tax numbers.


This shows that, for the 2011-9 period, corporate profits have not increased much and, in fact, have been quite flat for many sub-periods during 2011-9 (especially pre-tax). One has to juxtapose the GAAP corporate profits in order to detect any meaningful deviation and then the challenge is to go for explanations and see if it matters (for the ‘market’ and also for some specific corporate equity securities). Historically and typically, NIPA profits are about twice the S&P 500 profits.

I’ve made simple graphs on paper combining the two curves over the relevant periods but here are two helpful graphs for which deserved criticism is warranted but they show a few notable findings.



With the two curves juxtaposed, we see that the two lines have been very tightly correlated up to the mid 80s (and also tightly correlated up to the late 90s). The data also depicts an explanation for what could be identified as a cause for a deviation during the dot-com phase with the value of stock compensation mis-measured and under-reported during the build-up and the NIPA profits eventually revealing the true value of management’s stock compensation. A similar and likely larger phenomenon with a similar origin has been building up more recently. The top graph also shows the potential impact of revisions for the most recent years. The graphs above stop in 2019. Both GAAP and NIPA curves have been quite noisy during the 2020 exogenous downturn, during the centrally mandated artificial life support and during the reflated ‘recovery’ in 2021. Let’s keep the focus on the period from 2011-2 to end of 2019 for now.



Note: numbers for NIPA and S&P 500 GAAP are not the same; they are used for % change for the period. NIPA for the year is the sum of the 4 last quarters (after tax). GAAP are EPS operating earnings (after tax) as reported by Standard and Poors.

                                 End 2011         End 2012         End 2019         %change 11-19         %change 12-19

NIPA profits             1537.5              1821.5             1938.68                 +26.1                           +6.4   

GAAP                         96.44                 96.82               157.12                  +62.9                          +62.3

Note: NIPA pre-tax for the 2011-19 period:  +23.1%, for the 2012-19 period: +3.8%

Note: In order to quantify the difference between NIPA and S&P 500 corporate profits, what to choose for the starting year after the GFC recovery (to avoid the noise described above during downturns and recoveries)? i think it’s reasonable to take an average of profits of the two years (2011 and 2012) as starting points for the period ending at end of 2019, which comes to a 47.2% difference in growth between NIPA and GAAP profits. Note that revisions will be made for the more recent NIPA years and it’s possible or even likely that it will be established that more recent NIPA profits will even be lower if the underlying thesis of this post applies, given the relatively strong rise in share prices in 2019 and 2020. Those who follow federal income tax receipts will note that end of year receipts (received in July 2020 for the 2019 year and in May 2021 for the 2020 year) were quite large and likely correlated strongly to significant capital gains (some of which related to stock-compensation exercise). So the 47.2% number is likely a conservative number.



IMO, most of this (47.2%) difference comes from the value of options and stock awards that have not been fully recognized (as a cost of doing business) yet (and never will be under present rules) in the GAAP numbers.


Note that there has been noise in 2020 and 2021 and earnings plunged, recovered and more. Also, stock-compensation has mostly continued and buyback plans are back in vogue. Assuming the Q1 2021 S&P numbers are annualized for the entire 2021 year, year-end S&P operating earnings will have increased by 20.7% versus 2019 year-end numbers. Based on the inputs described in this post and others, I think it’s reasonable to assume that NIPA-S&P 500 disconnect has remained the same and possibly increased some more.

This means that, when assessing the market, for the period from 2011-2 to 2019 at least, it can be suggested that something like 47.2% of earnings are earnings above the NIPA corporate profit baseline and are basically earnings without taking into account unrecognized costs ie not true earnings. The relevant PE or CAPE numbers need to be adjusted to reflect this aspect. Using basing math results in the necessary reduction of 29% in GAAP earnings which results in a 1.41x adjusting factor for the adjusted PEs versus reported PEs. So this can be reasonably done for CAPE since the method involves adjusted earnings over a ten-year period very similar to the 2011-2019 period in duration and timing. The last reading for CAPE has been reported at 38.7 which means the adjusted for stock comp CAPE would be at 58. However, the CAPE ratio is based on 10 years and the disconnect has likely been building up gradually over the period and 58 is likely too high a number. Anyway this is sliced or adjusted, the CAPE adjusted for the disconnect has reached a level likely around 45 to 50, a level never reached before in the US and relatively comparable to what happened in Japan in the late 80s, taking into account accounting adjustments (whatever that means for the US now).


Potential limitations

The first objection is that the NIPA world is different (different companies, different sizes, different measurement methods etc) from the S&P 500 world. Maybe, but even if true, this did not seem to cause permanent, unexplained and irreversible deviations before, when stock options were not a significant and relevant variable.  Overall, the critical variable strongly appears to be the unrecognized stock-based compensation because there are no clear reasons as to why NIPA profits should suddenly, significantly and persistently deviate from the S&P 500 earnings and the stock-based compensation variable can strongly explain (timing, extent etc) the deviation based on observed inputs for many companies and in the aggregate. This assessment relies on a difference-in-differences principles method based on top to bottom consistent and rational connections.

Even if parts of the S&P 500 have, for whatever reason apart from the unrecognized stock compensation issue, diverged from the rest of the S&P 500 or from the larger world of investments, a fact remains that overall corporate profitability growth has been very poor and if there is any small segment that has done better, this means that the remaining larger segment has done worse than poorly.

A second objection is that the S&P 500 may contain a large fraction of companies which have a large component of foreign revenues and profits. Again maybe, but even if true, from official data, the increase in foreign profits (FRED corporate profits after tax, rest of the world) for the relevant period come to about 100B, an amount which is relatively significant in itself, but even if allocated entirely to the S&P 500 companies, would not materially explain, by itself, the growing deviation between NIPA and GAAP profits. Also note that foreign profits have been tightly correlated to foreign exports and the US entered an export recession starting in 2019 and which is still going on at this point.

A third objection is related to variable levels of corporate share compensation and share buyback activity. Again, while the level of such activity may vary and may be relatively concentrated in some sectors, the 47.2% calculated above is an average number for the market as a whole. Some companies will be more involved and others, less.

A fourth objection is related to the notion that there are a few large corporations which are dominant, have high margins and which are heavily involved in stock compensation and buyback. While this is true, compared to the dot-com period for example, share-based compensation is now widespread and, since the GFC, very high levels of corporate buyback activity have occurred widely. To illustrate this aspect, look at the differentiated market performance of the S&P 500 with and without its star or dominant performers.


Also, comparing the top 50 with the S&P 500:



Looking at this aspect from many angles and using many inputs, it’s possible that the S&P 500 group increased net profit margins for the relevant period by 1 to 3% versus the rest of the NIPA corporate world (thereby slightly increasing the gap that existed before between larger and smaller companies) but this number is very small versus the 47.2% estimate.

Also, despite huge buyback activity over the relevant period, overall, share count has been decreasing by less than 1% a year, a number that, even if compounded over the period, is small versus the 47.2% estimate.

The cost of business related to management compensation has not been completely recognized and this aspect essentially explains the NIPA-S&P 500 profit disconnect.

If you have companies in your portfolio with convertible debentures or such that need to be marked-to-market every quarter, just look at the huge impact that small change in assumptions make and the huge impact of rising share prices during the phase where conversion has not occurred. Even if you happen to think that this whole thing is bogus, it may be a useful exercise to try to determine the true cost of stock-based compensation in the companies making up your portfolios. It’s been an eye-opener for me at least. That’s what NIPA profits reporting accomplishes at the aggregate corporate level and the full cost of stock-based compensation can be evaluated for individual companies, even if not recognized, using GAAP reporting requirements in the footnotes.

But then again, some will say who cares?

Some would say it doesn’t matter since this not-yet-recorded compensation expense is paid by new shareholders joining on board along the way. To this objection, I would say not quite because i’m not sure if this cost has been truly integrated into valuations, along the way. Also, if the company buys back its shares on the open market and sits on the other side of the trade, calendar year free cash flows are used and, conceptually, the full price of this prior years’ compensation is assumed by the company itself, while bypassing the income statement, so already-on-board shareholders also shoulder these costs. If stock compensation management expenses are not expenses, then what are they? For example, the stock compensation packages are huge at Facebook and Microsoft but their share count has been decreasing very slowly lately, despite very significant buyback activity. Some management use the stock compensation to actually increase their equity stake in their respective companies but, in the main, equity obtained through compensation schemes are simply opportunistically sold with a short term orientation.

I wonder if this will not be looked at by regulators, eventually. In these types of scenarios, typically, these deficiencies are remediated after the fact.

And then what if this is already priced in by the market. Then fine. The only thing is that the PEs (as a rough guide to valuation) need to be adjusted to reflect these unrecognized costs. Using a rough one-third of earnings as unrecognized costs over total earnings including unrecognized costs (relatively conservative for the two examples), the PE for Facebook goes from 30, as reported, to 45, as real, and the PE for Microsoft goes from 39 to 58. I don’t know how to value both companies but for those who do, the stock compensation costs need to be recognized somehow. Also, PEs of 45 to 58 also result in earnings yields higher than the 10-yr Treasury, if that’s felt to be the relevant benchmark…

Maybe, the fundamentals warrant these true earnings multiples and history may eventually confirm that present prospects are truly outstanding but present (and adjusted) valuation levels are similar to the Nifty Fifty period (for the some of the leading issues) and to the very late 1980s in Japan (the whole market). In Japan, the outcome showed eventually that the lofty expectations were absolutely unrealistic. However, basic empathy requires to consider the Japan environment from the late 80s perspective which was quite optimistic about the future. The Japanese market was surfing on a wave which saw Japanese stocks return 23% per year over two decades (and an increasing trend for returns) and people simply expected more of the same. Their own late-80s PEs were quoted to be in the 60 range but an interesting feature was that accounting peculiarities of the Japanese-GAAP-equivalent made their PEs look higher than they really were, perhaps around 25% higher…Anyways their lofty valuation levels based on rear-view-mirror analysis met, head-on, with fundamentals that turned out to be unusually poor and enduring. Ouch!

For the Nifty-Fifty stocks, the historical record showed that the optimistic expectations were, at least partly warranted, for many of those stocks, but it took more than twenty years for the high-flying PE stocks to surpass the performance of stocks which had lower valuation multiples and one had to go through a few very miserable years (including the brutal 1974-5 years; these were inflation years, ouch!). Mr. Munger has talked about this period and how he had a hard time (he was not into the Nifty-Fifty frenzy but the whole market swooned), especially considering the other people’s money component. Examples of Nifty-Fifty stocks that ‘deserved’ to be held for the long term whatever their price (this does not apply to  many of them though) were (with high-water PEs and percentage decline in the nasty 70s): American Express-38  -75%, Anheuser-Busch-32  -70%, Coca-Cola-46  -78%, Disney-71  -86%, McDonald’s-71  -72%. A lot has been said about Mr. Buffett and the timing issue during that period and some think that he rolled over cash and bond holdings to equities during the 1974-5 downfall but that does not quite fit the reality. In general, during the corrective phase, the top 25 Nifty-Fifty stocks declined by 67% and the rest of the S&P 500 stocks declined by 29%. Entering the 1974 period, Mr. Buffett (through BRK) was highly invested in stocks that were not part of the Nifty ones (they were largely safe and even bond-like equities), waited for the recovery to sell them but took advantage of the 1974-5 period to back up the truck for some opportunistic equity opportunities using available liquidities. And the rest is history which often rhymes (He did eventually invest in KO).

Another consideration is that what has been happening in 2019, 2020 and 2021, related to this specific unrecognized stock-compensation issue, will likely be characterized (eventually recognized) by more of the same and more.


Final point

This unrecognized cost ties in well with what bizaro mentioned above about wage costs eating away at margins. If you simply look at GAAP profit margins these days for the period preceding the covid noise (and the associated corporate welfare, oups i mean corporate subsidies), you see a high and rising net margin trajectory which hides the gradually increasing rising trend in general wages (going from 2 to 4% per year, with no link to the 2016-20 term). If you add to that the presently unrecognized cost related to stock compensation (part of general wages but unrecognized), since 2012, you see a declining line, meaning that operating profits are going down in correlation to true and full operating costs and this divergence has been rewarded with higher multiples. Is that justified by low interest rates? A final consideration is that increasing share of free cash flows used for buyback is happening during a period of decreasing operating and net margins with higher use of corporate leverage and lower rates of capex investments, all aspects raising sustainability and resilience issues.



Take-home messages

-There is something weird going on in the cost recognition of stock-based compensation expense.

-The cost has been recently unrecognized and is giving rise to a large, growing and looking-quasi-permanent disconnect between basic core corporate profits and reported profits.

-This aspect (that needs to be adjusted) combined to general valuation levels means that we are going through one of the highest valuation period ever based on the multiple paid on core underlying earnings.

-This analysis has no value for prediction but consideration should be given to hold lower valuation securities or cash for optionality and self-administered intrinsic cognitive tests should be done using various stress tests in order to evaluate the meaning of long-term holding strategies.

-An interesting exercise for your favorite stocks, if applicable, is to actually go through the calculations necessary to estimate the true cost of the stock-based compensation arrangements (options and stock awards).

Link to comment
Share on other sites

Shiller has some very big holes in it ....


For Shiller to work, the vast majority of new share issuances has to be via the market (therefore at market price). Utter crap!.

Employee stock options do NOT settle at the market price on the day the option is excercized, they settle at strike prices settled YEARS earlier, and there are a LOT of these spread over a LOT of different companies in the index. Oops!


A great many acquistions are paid for via equity issuances that by-pass the market entirely (o/g). Immediatley income accretive, assets increase, equity increases, almost all ratios improve, yet the market recognizes none of of it at the time. Often worsened when the positive changes are so extreme that analysts dont trust their models, and therefore don't update untill quarterlies have been released. Oops!


Sure, when Shiller first published, the index may have been relevant - but that is getting on 20 YEARS ago.

Today it is a completely different world.



Link to comment
Share on other sites

I think the better way to look at SBC is to look at the dilution incurred relative to the amount of stock outstanding. Many SAAS stocks dilute quite a bit probably resulting in shares outstanding growing by or 3-5% annually, but this may still result in healthy per share growth, if a company is growing 30%+/ year. it is not so great for companies like WDAY where growth slows to 15% and yet shares outstanding still grow at a 3-5% annual clip.


In any case, the SBC that makes the GAAP income statement look so ugly is a result of the high stock valuations at a given dilution rate. That assumes however, that the companies won’t grant more SBC if the multiples drop. From my experience, this is generally the case, if valuation metrics’s drop, company won’t make up for this by issuing more stocks and options. You can tell this by looking at companies that have traded at high multiples in the past, but at row trading cheaper like CSCO, ORCL.


Ironically, even though they have aged gracefully in terms of morphing from growth into FCF plays, the market has not rewarded those companies much.

Link to comment
Share on other sites

^Shiller PE is only an input among many others.

"Sure, when Shiller first published, the index may have been relevant - but that is getting on 20 YEARS ago.

Today it is a completely different world."


Shiller PE has reached a permanently elevated plateau and this time is different. Maybe.



i've developed a different perspective (vs share-based compensation) when multiples drop: some of the compensation is lost but much is made up by granting new stock options and awards (at much lower prices). Companies also can use rules that allow them to re grant stock options or awards (replacing the previous batch) without losing the deferred tax asset as a result of the already recognized non-cash expense tied to the previous share-based compensation. i think most people do not realize how this expense has become a huge expense. There is nothing wrong with a large expense if the the future positive cash flows related to this 'investment' are worth it but (opinion) it's weird that a large part of this expense is not reported at the corporate level.


The case of Microsoft and the growth in various relevant rates.


Microsoft is different from the rest of the S&P 500 or NIPA crowd but it’s a useful example and it’s the one most closely linked to the SaaS theme.


The main differentiating factors for Microsoft have been significant increase in sales and margins, especially during the latter part of the 2011-2 to 2019 period, as well as large stock-based compensation packages and buyback programs. Overall, differences tend to cancel out for this analysis versus the typical stock and it’s a relevant and representative example of the NIPA-S&P 500 disconnect in corporate profits over the 2011-2 to 2019 period.


Evaluating the unrecognized GAAP cost eventually recognized by NIPA for a specific company requires some work, some estimates and some (retroactive) adjustments. Interestingly, GAAP requires companies to supply the numbers related to the market value of stock awards and options when the vesting date is reached. However, this is not to recognize or report the compensation as a true and full expense. It’s to allow the company to get a tax deduction for the true expense! When the true (but not reported fully) expense is higher than the cumulative non-cash amortized expense estimated upon the grant date, this results in a cash flow tax benefit in the year that stock awards or options vest. As mentioned, NIPA eventually captures this unrecognized cost and revises down prior years’ corporate profit results. So this true cost appears at the aggregate level but is never recognized for the specific company, Microsoft, in this specific example. This corporate cost not captured by the GAAP financial statements can be calculated by using information documented in the footnotes related to stock compensation and taxes, using the reported excess tax benefits per year and the yearly reported difference between the grant-date values of stock compensation and the vest-date values of stock compensation. There are potential sources of error. For example, the person exercising the option or actually obtaining the stock award may hold on for a while and the tax effect may be delayed in its recognition. Credible sources indicate that insiders’ sell/buy ratios have remained high and, looking at DEF 14As serially, it appears that stock compensation, overall, is not held for the long term apart from a core position built early on and rapidly sold when leaving the company. Also, the tax due on the sale of stock compensation may be mitigated by investments in professional sports teams or whatever but I assume that, overall, capital tax gains are rapidly paid by individuals benefiting from stock compensation at sale post vesting. There have been some changes in GAAP stock compensation reporting in 2016 but none of the changes impact materially the conclusions here.


For Microsoft, here are the assumptions and method used in order to derive the yearly unrecognized cost related to stock compensation:

-MSFT has moved away from options to stock award; stock awards vest after 4 to 5 years.

-For years 2011 to 2015, each year’s cost is determined by amortizing over the previous 5 years the “excess” cost determined five years after the said year. For example the cost for the 2011 year results from the addition of the yearly amortized cost determined in 2016 and amortized from 2012 to 2016, and etc for years 2012, 2013, 2014 and 2015.

-For the years 2016 to 2019, the same steps and principles are used and it is assumed that stock awards will continue to vest in a similar pattern for years 2021-4 as before ie with an estimation of forfeited numbers of stock awards based on the 2016-2020’s pattern. For year ending June 2021, the share price taken for vest-date value is 244, the mid-price for the year and for years 2022-4, the share price taken for vest-date value is 280 (more or less the share price now). In 2022-4, it’s possible that share price will double or more? but it’s also possible that share price decreases by 70% or more? so who knows? Obviously this assumption could have a significant impact for the last years of the period as the period from 2015 up to now has resulted, so far, in huge unrecognized cost because of the relatively high volume of stock awards in the preceding years granted at a much lower value than what was realized after.


Table (all numbers in billions, except D) and E) in %)

              A) unrec. cost     B) net income     C) sh.-based comp.expense     D) C/(C+A)     E) A/B

2011             0.893                   23.15                                 2.17                                 71%              3.9%

2012             1.197                    16.98                                 2.24                                65%              7.1%

2013             1.727                   21.86                                 2.41                                 58%              7.9%

2014             2.487                  22.07                                 2.45                                50%             11.3%

2015             3.261                   12.19                                  2.57                                44%             26.8%

2016             5.410                   20.54                                 2.67                                33%             26.3%

2017             8.249                   25.49                                3.27                                 28%             32.4%      

2018            10.651                   16.57                                 3.94                                27%              64.3%

2019            11.941                   39.24                                 4.65                                28%             30.4%



-A/B seems to compare apples to oranges tax-wise (pre-tax expense versus after-tax net income) but remember that GAAP rules allow the company to obtain the excess tax benefits in the year stock-based compensation reaches the vesting (or service) date for this unrecognized cost even if the actual excess cost is not recognized as an expense.

-In 2018, there were accounting changes and the impact of the Tax Act and net income was impacted by 13.7B so the adjusted A/B ratio is 35.2%.

-The table confirms that the unrecognized GAAP compensation cost (but eventually captured by NIPA) became very significant during the 2011-2 to 2019 period as the value eventually realized by stock awards granted at relatively low prices and then exercised 4 or 5 years later, when prices were much higher, was very much higher than the fair value estimated upon grant. For example, those who were recipients of the granted stock awards (83M shares) during the 2016 year at an average price of 41.51 and who did not forfeit these awards (my estimate is that about holders of 16M of those shares did forfeit) did very very well, even after tax. That’s fine and this post is not about a moral position on this. It’s just that financial statements should normally reflect, in a transparent way, the full cost of employee compensation.

-For the last 5 years of the period (2015-9), total net income, as reported, is 114.0B and the so far estimated unrecognized compensation cost is 39.5B (43.4B if the last share price last Friday is used on vest date ie 289.67 concerning years 2017, 2018 and 2019). This means that cumulative reported net income for those five years is not 114.0B but more likely around 71-75B.

-For the years 2016 to 2019 and likely up to now, it’s reasonable to suggest that, for Microsoft, net income should be reduced by about a third or more in order to obtain the true core earnings, taking into account the unrecognized stock-based compensation costs.

-During the 2011-2 to 2019 period, EPS (as reported) compounded (per year) at 10.0%, share-based compensation as reported at also 10.0% and unrecognized compensation cost, as estimated, at 23.5%.

-On a net basis, for the same period, shares outstanding, diluted, compounded (per year) at -1.3% for the period and only at -1.0 to -1.1% during the latter part of the period, indicating that previous MSFT shareholders were, in large part and through financing cash flows, on the other side of the trade when the NIPA/S&P 500 corporate profits disconnect reached high levels and, essentially, buying shares at market prices way higher than grant-date value previously allocated from the market price contemporary to the stock awarded 4 to 5 years before. In other words, for example, MSFT is buying back stock now, in parallel to just vested stock, at around 280-290 in parallel to awarded and granted stock 4 to 5 years ago at 41.51 to 55.64. Now, that doesn’t necessarily mean that MSFT is doing a bad deal when buying back its stock these days but if the primary motivation to buy back stock is to keep share count more or less constant, then one has to wonder about the value proposition.

-It’s also interesting to note that, for the same period, insiders’ percentage of stock ownership went from 10.42% to insignificant (1.39% in 2019 to below 0.05% in 2020) as Mr. Ballmer disappeared in 2014 and Mr. Gates decreased his personal holdings progressively. So, looking also at the present management team and the pattern of stock ownership over time, it’s reasonable to suggest that the huge stock-based compensation allocated during the last ten years was not a buy and hold for the long term type of decision.

For the 4 years 2011-2014, average cash flows allocated to buybacks were 7.3B per year or 24% of average cash flows from operations. For the 5 years 2015-2019, average cash flows allocated to buybacks were 14.5B per year or 37% of average cash flows from operations.

-For Microsoft, today’s PE of 39 is, in fact, more like 58-59. Not bad for a “SaaS” company’s market cap of 2.18T.

-Doing this exercise for a sample of other companies (SaaS and non-SaaS) yields results that strongly suggest that the NIPA-S&P 500 corporate profits’ disconnect is essentially a story of unrecognized share-based compensation costs over the period.

Link to comment
Share on other sites

GOOG just grew 62% YoY...what's the right Shiller for that?


It's at <30x 2022 estimates (which will probably be revised up), and that's after being up 60% YTD. 


pretty incredible stuff. 


Edited by thepupil
Link to comment
Share on other sites

On 7/27/2021 at 5:31 PM, thepupil said:



It is amazing to see how robust GOOG has been through this entire pandemic given that the travel industry was one of their biggest, if not the biggest, advertiser. Its mind boggling to see their results up so massively knowing that there has been very real, very long-term damage done to that sector. 


I bought Google in the bust of 2009 and rode it up and was very happy to do so. I thought about doing it again this time around, but couldn't get over my concerns that there would be moderate impairment in both their advertising customer base and the consumers who consume the advertisements. It seems I was very clearly wrong. 

Link to comment
Share on other sites

  • 3 weeks later...

^A few comments about Google (a company about which i cannot form an adequate or relevant opinion concerning business prospects or valuation) and about the potential relevance of reversion to the mean for concepts such as CAPE.


After looking at GOOG's annual reports for the idea developed above related to the unrecognized cost of share-based compensation (this is a more difficult task for Google compared to Microsoft for various reasons related to GOOG's policies and disclosures and the company withholds a percentage of vested shares to pay the income tax liability of the 'employee', related to this compensation). Assuming share price remains at around 2700 over the next 4 to 5 years and taking into account that share count has been reduced by only about 0.3-04% per year in the 2015-2020 period, this means that, for the 2015-2019 5yr-period, 75B of compensation costs will not have been recognized, compared to cumulative 114B net income for the same period. It's not 'fair' to compare this way because the 75B unrecognized compensation cost should be matched to the cumulative net income over the 2020-2024 period. However, if net income continues to grow at the same rate, it would be reasonable to expect matching share price growth. For GOOG in the future and only looking at unrecognized share-based compensation arising as a result of RSU granting activity during 2015-2019 period means an incremental 40B of unrecognized compensation cost per 1000$ per share price increase on top of the 75B mentioned above. It's reasonable to suggest that GOOG's net income has, is and will not take into account that reported net income does not factor in a 30-35% reduction in net income per year if unrecognized compensation costs would be recognized.

All that to say that GOOG may be a great deal now and massive buybacks may make a ton of sense (i truly don't know). Without a 'moral' or regulatory perspective, it just means that GOOG's PE is not 27-28 but more like 39-40. That's all.




The CAPE topic is mostly remotely interesting or relevant but sometimes the math gets challenging (basic math). One of the underlying questions is to what extent reversion to the mean applies or not (range from 0-100%).

When factoring in reversion to the mean, especially for segments of the companies followed and even more so for individual names, there are two basic problems:

1-reversion to the mean may NOT (like in never) occur.

2-reversion to the mean may take a long time to manifest, the path to the mean (and beyond) may be irregular and a long muddle-through phase may smooth out potentially interesting fluctuations in price.

i think 1- is the possibility that Mr. Shiller described recently versus low interest rates. Even if it's hard to correlate general levels of interest rates to valuation levels in general and returns achieved, it's possible that, for conditions now (ie the last 20 years or so), interest rates and the yield differential with equities may play a fundamental and driving role in valuations. i think some people call this non-stationary of conditions. And with negative interest rates in sight (opinion) who knows?

2- is the reason why people who are right may look like clowns (there's also the possibly that clowns are wrong).


One of the defining features of the modern era is that size no longer matters the way it did and one has to decide if that's also reached a non-stationarity situation. Mr. Buffett, early on in his partnership letters, described this potential drag on returns in relation to his own size (portfolio). These days, it seems that even the opposite is true ie as many firms grow larger, they become even more resilient and have somehow developed the capacity to dominate even more. Like ultra low interest rates, the size phenomenon has been quite a game changer for a while now. The size phenomenon is quite an unusual historical development.


When i was born and after, people started to question the US' economic ability to grow as it was getting very large.  If one looks at real GDP growth over certain time periods (for fun this is compared to real corporate earnings growth {Shiller data}), the size didn't appear to matter for such a long time.


                                                           1880-1920          1920-1960          1960-2000

real GDP growth (%)                               1.4%                    1.8%                     2.3%

real earnings growth (%)                        0.3%                    2.0%                    2.3%


For the 2000-2020 period, real GDP growth came at 1.2% and earnings growth came at 3.4%. Earnings growth are projected to increase by about 50% in the next two years which should help (if it happens) for the irreversion (the word irreversion does not truly exist but irreverence is the closest synonym if mean reversion is a thing to respect) to the mean for PE. 


Also, it used to be very bad news for large companies entering the Fortune 50 select club:



No more, it seems. Reading a very interesting report lately discussing the growth potential of the select few: FANG, FAAMG etc, and applying the underlying assumptions to the greater picture suggested that the revenues only from those few would result in a revenue number larger 100% of GDP in a reasonably foreseeable future. The math eventually gets challenging even if negative interest rates are used.


Recently, i also saw this well done survey about expectations (investors with a 10-yr horizon:


Using the above numbers for where we're at in 2021 and integrating into quite reasonable assumptions for the real economy and general investment inputs, it is then expected that PEs in 10 years will be at 72. If one integrates the unrecognized compensation cost described above, 'we' should then reasonably expect a true adjusted market PE of 100. The bigger the better?


Anyways, in his 1999 general valuation article, Mr. Buffett noted that, for lofty expectations to be met, three things needed to happen: 1-interest rates getting lower (getting lower, they did to the relative dismay of the Fortune article author over time), 2-corporate profit margins as a % of the economy getting larger (getting larger, they did despite the expectations of growing divisiveness if they did) {Note: NIPA corporate profit margins did decline significantly during the 2011-19 period but were still way higher than in 1999}, and 3-if a certain investing mentality persisted.

Mr. Buffett is wise and did not make any predictions or forecasts then. He only said that the math was getting challenging and that expectations should be moderated.

i wonder these days if it's the math that's getting incredibly challenging or if i'm not one who's mentally challenged.

Link to comment
Share on other sites

@Cigarbutt, I'm not really sure I get your point on stock based comp. 


If sales/share, EBITDA/share, Cash flow/share, free cash flow / share, net income/share (<--i get that you're saying net income is overstated) are all up and to the right, then I don't understand why should penalize companies for when the stock goes up over some time frame such that issuance to employees is with hindsight more expensive. 


by saying "Google is actually 40x earnings" aren't you penalizing earnings for FUTURE stock outsized stock price appreciation? 


I've read your stuff on this twice and still don't understand. 



Edited by thepupil
Link to comment
Share on other sites

53 minutes ago, thepupil said:

@Cigarbutt, I'm not really sure I get your point on stock based comp. 


If sales/share, EBITDA/share, Cash flow/share, free cash flow / share, net income/share (<--i get that you're saying net income is overstated) are all up and to the right, then I don't understand why should penalize companies for when the stock goes up over some time frame such that issuance to employees is with hindsight more expensive. 


by saying "Google is actually 40x earnings" aren't you penalizing earnings for FUTURE stock outsized stock price appreciation? 


I've read your stuff on this twice and still don't understand. 




I'm also curious about this view.  I had a similar discussion in the NVR thread (see March 17, 2020 posts): 


The ex post, hindsight method of finding the "true" cost of stock comp seems to contain a paradox:  The more the stock price goes up (and thus the more equityholders have benefited) the more they appear to have been robbed by "excess" employee comp. 


Link to comment
Share on other sites

^Short version:

A similar debate occurred in the late 1990s and early 2000s about share-based compensation and, obviously, there were/are valid and different perspectives but the challenge, again, consists in answering the following question: If share-based compensation is not an expense, then what is it?


Longer version:

It's a multi-variable question so let's unpack. What this 'idea' is NOT about:

-Is share-based compensation right or wrong?

-Is share-based compensation an appropriate way to reward employees?

-And even: Is the buyback decision a good one (capital allocation and valuation aspects)?


If you can spot an investing idea (business prospects and valuation) and if the compensation/repurchase activities make sense, the thinking activity for this is irrelevant and can stop.


The issue is that this 'temporary' unrecognized compensation cost has become very large, has been going on for about a decade now and, overall, the buyback activity has essentially (not quite) balanced the share issue from compensation to employees, making this cash outlay, conceptually, an operating expense that is not recognized.


In May 2019, Mr. Yardeni released a piece (the drive was political ie don't fight buybacks but the data reported helped with the recognition issue) basically arguing that huge buybacks since 2011 had not really contributed in a big way to the growth of EPS because: "Since the first quarter of 2011, a total of 72 billion shares were repurchased. However, over that very same period, the number of outstanding shares declined by only 22 billion!". Viewed this way, overall, stock buyback activity appears to be matched to share issue related to share-based compensation with the goal to keep share count more or less constant.


The issue is that there has been a HUGE growth (absolute and relative) in the difference between grant-date value and vest-date value of share-based compensation. This value can be measured. It's clear who gets the value but it's unclear who pays for it although it has to be the equity holders, somehow.


The best analogy or concept i could use to extract the meaning of this disconnect is the use of loyalty points at credit card companies. It is clear that the credit card user (maybe the merchant) is the ultimate payer, in the aggregate, for the use of loyalty points. In the aggregate, you are paying the toaster that, supposedly, you are receiving as a ‘gift’. Credit card companies (and others like airlines) have had the genius to devise this free-product mentality. Whereas there are some people who clearly benefit from the judicious use of loyalty points (i think thepupil and KJP would be such people), this is a zero sum game in the aggregate (clients are ultimately paying for their gifts) and there is a bunch of patsies at the table. Somehow, the corporate world has orchestrated (voluntary, involuntary or whatever), or reacted to, a situation where the full cost (most of it) of share-compensation expense has become huge and is not recognized but is still nevertheless paid by investors who don't see it and who may happen to think that it's a free lunch. 


If share-based compensation is not an expense, then what is it?

Link to comment
Share on other sites

3 hours ago, Cigarbutt said:

but the challenge, again, consists in answering the following question: If share-based compensation is not an expense, then what is it?


I agree it's an expense. how one should account for it is up to interpretation.  


how would you adjust GOOG's earnings (or market cap) on a forward looking basis to account for what you see as a systematic undercounting of stock based compensation expense?


the sell side seems to add back stock comp expense, then builds in an estimate of shares outstanding (which will include the impact of RSU's). this does not seem overly aggressive to me. 


Said differently, JP Morgan sees Google doing $69 billion of GAAP Net income for 2021 and $76 billion of "adjusted net income" the biggest add back to which is $15 billion of stock based comp for 2021. I've snapshotted their model which shows you the SBC as an addback.  JP Morgan is not saying "stock based comp isn't an expense" by adding it back. Rather, I think think they're baking in that expense by increasing the share count. 


Goog seems to consistently have 18-22 million of unvested shares from RSU's at any given time 2015 to present.. This is 2.7-3% of outstanding shares and they grant about 1-2%/year to employees. The share count was roughly flat 2017-9 and is now on a trajectory of -2%/yr as the gross buybacks have increased. 


 It's true that the 12.0-13  million shares that were granted / vested in 2015 are worth $32 billion today, which is a lot more than was expensed for those shares. 

But I fail to connect that fact to my (or anyone's) assessment of what Google is trading at or might be worth. If you account for it by increasing shares (or rather not decreasing shares by the full amount of the buyback) then you've accounted for it, right? I feel like I'm missing something here. Your takeaway seems to be that GOOG (and companies of its ilk) are overreporting earnigns power by a substantial margin and that's not at all my takeaway from this. 


I think you might respond "what if they had to pay cash instead of grant RSU's wouldn't GOOG then report much lower income?". In theory that could change things, but in practice, it's hard to imagine how or why that may occur (and they are paying cash for that expense via the buyback). 


Overall, I think I understand what you are saying, but you seem to be making a leap that megacap tech (and the market as a whole) are making significantly less money than reported and are thus more expensive because of this. The only way I see that being true is if their collective growth suddenly slowed, their stocks fell AND you had to keep the same dollar amount of stock based comp (ie you had to grant a higher % of shares, ie it'd grow more dilutive over time). I see no evidence of that and don't have that view. It's a possibility, but not the base case, in my opinion. 



Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

  • Create New...