Two of the most intense words on Wall Street may be "earnings" and "quality." For the most part, we hear little about this linguistic pairing, but when we do, especially when it comes in the context of an accounting controversy, you can usually settle in and reach for your popcorn in anticipation of a really big show (and if you happen to have a position in such a stock, make sure your seatbelt is very securely fastened). But what about the plain vanilla day-to-day world of investing. Absent a high-profile announcement (e.g. a short attack by a prominent hedge fund - that sort of thing), does anybody really care?
Clearly, we should care. There are many perfectly legal things companies can do to make reported earnings look better and if we're as interested in getting things right as many of us say we are, we really ought to prefer Company A, which reports earnings of $2.00 per share using simple clear-cut accounting policies in lieu of Company B, which reports $2.00 per share but pretty-much requires you to spend hours studying accounting textbooks in order to figure out how it reached the number. This is not just a matter of simplicity. Chances are Company A's $2.00-per-share consists mainly of cash while Company B's results may contain a considerably smaller cash component and depend heavily on accounting "accruals" to reach the final result. There, I said it: the "A" word. If you care about earnings quality, get used to it.
All about the A-word
An accrual is a non-cash event that impacts the financial statements. Here's one very simple example. I sell you a machine. You take delivery immediately and start using it (i.e. there's no doubt this is a legitimate transaction), but don't pay for it right away; we agree that you'll pay within 90 days. I record the sale in my books. On the income statement (which keeps track of my ongoing financial progress), this is recorded under, you guessed it, Sales. I also have to record it on the balance sheet, which keeps tracks of what I have and what I owe. You'd think that if I sold something, I'd have cash and be able to record that on my balance sheet, which is what McDonald's does when I hand over cash in exchange for a burger. But the machine is a bigger transaction. As with many transactions that take place every day in the business world, I don't have the cash that corresponds to the sale. Instead, I have an I.O.U. So I record the value of this obligation under Accounts Receivable (which, as a matter of custom, is usually located on the balance sheet right below Cash). So there we have it. I made a sale (and logged it on the income statement) and I recorded what I got from the sale (an I.O.U.) on my balance sheet. Note, though, that the I.O.U. is not the same thing as cash. Therefore, it's known as an accrual. When you pay your bill hopefully as agreed within three months, I'll subtract the amount from Accounts Receivable and add it to Cash.
Company earnings come in two parts: cash plus accruals.
If you read enough earnings-quality literature, you may find yourself thinking that accruals are evil. They aren't. They're a natural and inevitable part of modern commerce. If you don't believe it, close all your checking accounts and cut up all your charge cards and see how long you can go without raising the white flag. (However bad it may be as a consumer, it gets far worse if you're in business and need to build new production capacity.) And by the way, you'll also need to insist that your employers, clients, customers etc. play along and have a suitcase full of cash handy in order to deal with you. (Good luck with that!)
Generally, this is fine. People and businesses typically pay their obligations as they come due. Yes, we all heard about the sub-prime mortgage crisis and receivables disasters such as those that befell the circa-2000 nouvelle telecomm firms. But think about why we heard about them? It's because they were news. And what, to a member of the media, is news? The adage they're taught is that if dog bites man, it's not news but if man bites dog, that's news. You only hear about such events because they are unusual. (How irritating would life be if CNN were to cut into regularly scheduled programming to announce that so-and-so paid an invoice for such-and-such!)
In terms of earnings quality, there are two issues with accruals: First, there is the possibility they won't eventually be converted to cash as we expect. In terms of the economy, one company in several thousand that finds all or most of its receivables uncollectable is a tolerable misfortune. But if you own shares in that company, it would be a disaster. The other issue is that companies can't do the kinds of things with accruals they can do with cash. Unless they're willing to amp up business risk by borrowing more, they can't use accruals to fund research and development, implement capital spending programs, improve marketing and distribution, pay dividends, make acquisitions (even sensible ones), repurchase shares, etc. So as investors, while we recognize that accruals are an inevitable aspect of commercial reality, all else being equal, we'd rather see the companies we own have bigger portions of their earnings be in the form of cash rather than accruals, or put another way, have higher earnings quality. (Imagine a discounted-cash-flow geek/value investor standing up at an annual meeting demanding that management liquidate and distribute the company's accruals to shareholders.)
Are you getting annoyed with how long it's taking me to get to Amazon.com (AMZN) and/or Netflix (NFLX)? I get it. After all, that probably is why you clicked on this article. I'm not quite ready yet, but if it'll help, I'll give you a quick preview. Accruals can also be negative, which means a company is taking in more cash than conventionally reported earnings suggest. That's so with AMZN and NFLX and it's something we should consider as we contemplate those gravity-defying P/Es so many rant, scream, whine, cry, bleat and curse about. Don't assume I'm going to reach a bullish conclusion, though; I've got to hold something back to get you to keep reading.
Does the Stock Market Care About Earnings Quality?
The answer here is a definite maybe.
The big-name in accruals research is Richard Sloan and another academician, Messod Beneish made quite a stir with his work in using ordinary data from SEC filings to try to identify companies more likely to use aggressive accounting to manipulate earnings. Both produced interesting research suggesting that investors could profit from consideration of such factors, but both worked a while ago and conventional wisdom today holds that enough investors caught on to have caused much excess return to be largely arbitraged away. In other words; great stuff back in the day but less great today.
I've been working on developing some earnings-quality factors to add to portfolio123.com and my initial testing of a variety of items confirms that none of them has the sort of silver-bullet power that would simplify our lives. But I don't buy into the notion that all the benefits of earnings-quality analysis have been arbitraged out of existence. We just have to be more creative in how we use these factors on their own and in combination with other ideas.
Calculating accruals is a bit of an art; there are several different formulations each of which provides its own unique answer. I'll be working here with accruals as defined by Richard Sloan (see Appendix at the bottom for details). I built a simple portfolio123 screen looking for non-financial S&P 500 constituents having the five lowest accruals-to-operating profits ratios for the trailing 12 months. Figure 1 shows the results of a one-year backtest (assuming rebalancing every four weeks, which is probably more frequent than I needed since this is a low-turnover model).
Figure 1 - Highest Earnings Quality Firms
Figure 2 flips the strategy and selects companies having the five highest (supposedly lowest quality and presumed most bearish) accruals-to-operating profits ratios.
Figure 2 - Lowest Earnings Quality Firms
Investing in the low-earnings-quality companies would not have put you into the poor house, but the test indicates significantly lesser performance than would have been the case had you focused only on the firms with the highest earnings quality.
Have you figured out yet which list AMZN and NFLX wound up on? Before considering those companies, let's recognize that consideration of five-stock groupings examined over the course of just 12 months isn't going to impress anyone in terms of sampling. So let's stretch it. Figures 3 and 4 look at the best and worst 25 earnings-quality firms respectively and run the tests from 1/2/99 through the present.
Figure 3 - Highest Earnings Quality Firms
Figure 4 - Lowest Earnings Quality Firms
I suppose that means we can shelve the sampling objections.
More importantly, bear in mind that these tests were run only against non-financial S&P 500 companies. (If you want financial firms, more homework needs to be done since the most well-known earnings-quality metrics are based on non-financial reporting.) This is huge. The S&P 500 is by far the most "efficient" part of the market; i.e. the area where it's hardest to come up with models that are able to identify potential winners. Needless to say, the better earnings-quality group also walloped the less group when I expanded the sample to include non-financial Russell 3000 constituents and interestingly, with this group, the performance strength was most pronounced in the more recent years (the period when many had presumed the benefits of earnings-quality analysis had been arbitraged away).
Good, we get it - Now talk about AMZN and NFLX!
OK. I hear you.
First things first: As of this writing, AMZN and NFLX are on the good (high earnings quality) version of the five-stock screen. That's clearly going to upset and enrage many readers who for whatever set of factors upon which they fixated have convinced themselves that Jeff Bezos and Reed Hastings are public-enemies numbers one and two. But according to the earnings quality formulation created by Richard Sloan, one of the most respected names in this field, if not the most respected, AMZN and NFLX both come out pretty good by virtue of accruals that are significantly negative, meaning their reported EPS figures fail to hint at the cash-generating aspects of their respective businesses.
With NFLX, we're dealing with a lot of fancy unfamiliar jargon but deep down, much of what's going on comes right out of Accounting 101; big expenditures expected to produce benefits over multiple accounting periods are depreciated (amortized in the NFLX language) over the life of the purchased asset. There's nothing exotic here. Companies do it all the time as well they should.
If a company that normally produces, say, $25 million a year in operating profit decides to spend $250 million to build a new plant expected to boost revenues over the next 20 years, subtracting the $250 million all at once would not give a proper picture of the company's economic performance. Undoubtedly, the cash-flow police will jump up and object: "Cash is King! No accruals!!" In year one, they would indeed, look quite macho. "We're going to show that damned company. No slimy CFO is going to pull the wool over our eyes!" Acceding to pressure from the earnings-quality police, the company posts a year-one loss of $225 million and financial message boards get lit up by shorts screaming about how the stock should tank because the company is so deep in the red.
But what happens in year two? Say the plant helps the company generated $10 million in additional revenue; so now operating profit is reporting $35 million. Wait a minute! That extra $10 million was revenue. What about costs associated with this new business? The slime-ball CFO says: "Hey guys, you broke my will last year. I did what you wanted. I restated year-one earnings to eliminate depreciation and I expensed the entire plant up front. So now, in year two, I get to book the extra revenue (which is quite real; it's all right here in this suitcase; see how neatly I stacked the bills). But the expenses associated with these additional revenues are now zero because it was all booked last year. And better still, because I listened to you, I'll get to inflate my reported profits not just in year two, but again in year three and again in year four and so on through year 20! So tell me again about how vigilant you guys are. Chuckle, chuckle."
With NFLX, there are lots of little accruals all over the place (as there are with most companies). But the big item that drives the Sloan formula is the backing out of depreciation and (more importantly for NFLX) amortization (of the content it licenses for streaming). There is, of course, a separate issue for NFLX and all other companies. Can it pay out the money it needs to fund its big off-income-statement projects? Sometimes, that can be troublesome, but NFLX has more than enough cash and borrowing power to handle it. So we can afford to keep our focus on measuring annual economic performance; i.e. the income statement plus whatever adjustments we feel may be warranted.
So how might we re-think NFLX's horrifying P/E (about 503, yes, that's 503, not 5.03 or 50.3).
We know that because of negative accruals, NFLX's reported EPS does not give a good picture of its economic performance. Let's start by adding back depreciation and amortization. That brings trailing 12 month "cash flow" up to about $31 per share and produces a price-to-cash flow ratio of about 6.6, which not only is not stratospheric but downright cheap.
Let's not jump so fast. Textbook cash flow, net income plus depreciation and amortization, is not necessarily a great metric. To improve it, we really need to subtract capital spending, or at least maintenance capital spending (doing so approximates Warren Buffett's preferred metric, which he refers to as "owner earnings"). For NFLX, capital spending is pretty low so that wouldn't alter the valuation metric. But having read my Graham & Dodd and recognizing that analysts must sometimes reclassify things in order to get numbers that make economic sense, I'm going to pretend an item from the cash-from-operating-activities section of the cash flow statement was classified as a capital expense: "Additions to Streaming Content Library." It's not a capital expenditure because when accounting conventions were adopted, everybody was thinking about the manufacture of widgets and not psychically envisioning the future NFLX business. But if you really think about it, money spent to acquire streaming content licenses serves, for NFLX, the same function capital expenditures would serve for conventional manufacturers.
Now, here's where things get a bit dicey. For the last couple of years, content addition costs have been huge - beyond huge, massive. If we subtract them as is, we're right back to square one; worse actually since trailing 12 months streaming content acquisition costs exceeded depreciation and amortization. So the simplistic for-dummies analysis remains as it was; NFLX looks frighteningly overvalued. But to really understand the market, one really needs to go beyond a for-dummies level of analysis. Surely you don't think Warren Buffett got to be Warren Buffett by memorizing simple formulas and mindlessly plugging in whatever numbers he saw in the company reports? He didn't subtract capex as it appears in 10-Ks and 10-Qs; he figures out the amount of spending "the business requires to fully maintain its long-term competitive position and its unit volume" (1986 shareholder's letter).
So what, exactly, should we do with NFLX? The big issues here are I) whether the current levels of content acquisition costs are likely to be sustained and ii) how much additional sales and operating profit this additional content can help NFLX generate. For the first question, we can presume NFLX pretty much finished spending to build up its basic streaming library as it transitioned from a DVD-mailing company to one that emphasizes streaming and wishes deep down DVDs would vanish. On the other hand, NFLX is getting into the production of original content.
In calendar year 2011 and 2012, additions to the streaming library amounted to $2.3 and $2.5 billion respectively. In 2010, it was $406 million. In the years before that, it bounced around in the vicinity of $100 million.
In the first quarter of 2013, money spent on additions to the streaming library dropped 22%. On the latest conference call, when asked in the most recent quarterly figure ($592 million) represented a good run rate, the CFO responded: "I'm not sure how to answer a good run rate on content acquisition. I would say that we continue to expand our content. And, you'll see that expansion has declined over the last year and two years. So, the rate of expansion has started to slow but we'll continue to invest in content." Later on the call, management confirmed that no first-quarter costs were deferred into future quarters.
I've seen Netflix criticized for not providing as much information as investors might like. This is arguably a valid complaint, although many other companies are similarly reticent, and it would be understandable if one wanted to avoid NFLX on this basis alone. For those who want to go further, let's say the ongoing run rate for annual spending in this regard settles in around $1.5 billion with no discernible increase in operating profits. Treating that as the functional equivalent of capital spending would, when combined with conventional capital spending, result in a multiple of about 46 times owner earnings per share. That's still very high, but not anywhere near 500. Also, it still assumes a heck of a lot of content spending and no incremental revenue. If we assume $1 billion in ongoing annual content acquisition costs, still with no additional revenue, the price-to-owner earnings ratio drops to 16.
This does not by any means make NFLX a buy. This is a complex business (the accounting is such because we are accustomed to reporting rules designed for widget manufacturers) and there are other issues: How effective will NFLX be in being able to continue to obtain content? What about subscription trends? What about pricing, subscriber acquisition costs, etc. But one thing is clear. It would be naïve to assume Mr. Market is as out-of-it in valuing NFLX as many believe. The key to valuing NFLX begins with an understanding of the way accruals correct the distortedly low picture of ongoing economic performance we get from conventional reported earnings.
In a purely accounting sense, AMZN is easier to assess than NFLX. Like NFLX, it has huge depreciation and amortization accruals that must be added back in order to compute cash flows and owner earnings. Capital spending is recorded in a more conventional manner; on a line labeled capital spending. This is because the money is being spent pretty much the way accounting standards presumed it would be spent; on "stuff." The stuff may consist of sleek looking servers and the like, rather than noisy dirty factory equipment. But stuff is still stuff and poses no accounting challenges. The other category of Amazon's spending surge is likewise conventional; marketing-oriented outlays whether in the form of more people, loss-leader Kindles (I'm assuming there is a loss per unit sold), aggressive pricing. The key is that it all runs through the income statement just the way analogous expenses would in a widget company.
Just to refresh ourselves on accruals, we note that one category of future-growth spending is subtracted from sales on the income statement and another is handled as a non-cash accrual. The difference is not based on what the outlay is being used for (the goals are the same) but on whether the outlays are likely to impact profits over the course of a single accounting period (price reductions, Kindle losses) or over the course of multiple periods. As discussed in connection with NFLX, were we to deduct all of the latter immediately, as the cash-is-king police might demand, AMZN's results would look much worse in period one and then be distorted upward in many subsequent periods due to an artificial assumption that such costs amounted to zero.
There is one other factor for AMZN, unearned subscription revenue, mainly for Amazon Prime and Amazon Web Services. If a subscriber pays up front for a year, AMZN has the cash right away and gets to do all the things companies can do with cash. But it won't all go on the income statement. It'll be allocated to the income statement over the life of the subscription (i.e. added one step at a time to revenues and subtracted one step at a time from deferred revenues). This isn't a big deal for NFLX since its subscriptions are monthly, meaning there's no need to defer subscription revenue from one period to the next. (That may change if NFLX starts pushing and customers accept annual plans.) The key is that these deferrals, like depreciation, are negative accruals, meaning they must be added back into reported income to get a sense of how the company is doing in generating cash. Actually, though, I'm not going to address these further right now because on the question of AMZN stock valuation, the impact of this is dwarfed by the impact of depreciation.
The basic valuation computations start out easily, as they did with NFLX. Trailing 12 month net income for AMZN plus depreciation and amortization amounted to cash flow of 6.30 per share suggesting a price-to-cash flow ratio of 41, still very high but not in the hundreds. And like NFLX, when we subtract trailing 12 month capital spending to approximate owner earnings, we get a negative number. And again like NFLX, the current capital spending number is exceptionally high, way, way, way out of proportion to historic trends. Trailing 12 months capex amounted to $4.069 billion. Over the past 10 annual periods, from 2003 through 2012, they came in at $46 million, $89.1 million, $204 million, $216 million, $224 million, $333 million, $373 million, $979 million, $1.811 billion, and $3.875 billion respectively.
If we were at or near peak levels, as might be the case with NFLX, we could plug in some assumptions and make a quick owner-earnings valuation case. But even if AMZN's capex run rate were to sink down to $500 million, that would still imply a 50 price-to-owner earnings ratio. The "problem" is that we are in no position to plug in such an assumption, even on a for-fun basis. Capex at AMZN is still rising and AMZN's 10-K commentary makes it clear it will remain elevated for a while.
So even with adjusting reported results for the big negative accrual, you still can't make a real valuation case for AMZN. That doesn't mean the stock is a short, as many believe. Value is not the only game in town. And what AMZN is doing, spending big to generate bigger revenues in the future, is not at all unprecedented. When considering precedent, many assess AMZN in terms of late-1990s new-economy companies, which is understandable since AMZN started out as one of them. Actually, though, if one spends less time reading hysterical rants and more time reading 10-Ks, it becomes apparent that cable TV and wireless might be more useful comparables. They, too, spent massively to build infrastructures and did so many years ahead of revenues. In fact, unlike AMZN, those firms used very heavy debt financing and were very deep in the red based on net income and even cash flow (and much more so based on owner earnings). And over the years, Mr. Market has proven himself quite adept at distinguishing between red ink that signals a troubled business versus red ink that is prelude to better things in the future. If you want to make such distinctions, step one is to dig into those 10-Ks and understand what the numbers actually are (as opposed to what the Bezos-sucks fanatics say they are); be particularly attentive to the presence of accruals (negative accruals in the case of AMZN), and think analytically about why they are what they are.
Back to the Accruals Screen
Are NFLX and/or AMZN for you?
Maybe they are. That's fine. These companies are hardly obscure and all the numbers are out there and well analyzed by many investors, so if you're into either or both of these stocks, you have plenty of good company. I tried to provide a framework that will help you understand the approaches the bulls are taking.
Maybe they aren't for you. That's fine, too. Not everybody is into every kind of stock. I'm into stories like these but when it comes to my own money, I prefer simpler stories that arise in a less-crowded space (micro caps, or actually, sub-micro caps). That's why I don't presently have a position in either AMZN or NFLX.
That said, I am into model-based investing and I have to say I was surprised at the performance characteristics of the simple one I built based on Sloan accruals. My work in this area is ongoing, but if I ultimately do settle in on one or more models based on this theme, it is entirely possible I might at some point be long AMZN or NFLX based on a model. (One thing in particular I'll be looking into is whether performance can be improved by adding in consideration of owner-earnings based valuation. I also suspect there may be merit to an approach that looks at an ordinary market-matching universe of companies and instead of buying or overweighting high earnings quality, underweighting or eliminating the few with the lowest earnings quality. These are topics for another day.)
Here's the way Richard Sloan calculates accruals.
ChangeCOA - Change COL - Depreciation and Amortization
ChangeCOA: Change in Current Operating Assets, which is Change in total Current Assets minus Change in Cash & Equivalents
ChangeCOL: Change in Current Operating Liabilities, which is Change in total Current Liabilities minus Change in Short term debt (i.e. any kind of debt included in the current liabilities portion of the balance sheet) minus change in taxes payable
Note: In his research, Sloan scaled accruals and net income by total assets. Others use different scaling factors. I chose Operating Income after Depreciation but will be exploring the other approaches as I go along.