The Components of Quality

PDF Version: P123 Strategy Design Topic 4C – The Components of Quality

If you’re ready to walk in the shoes of Warren Buffett et. al., this is a Topic that’s very important, perhaps even more so than Value per se. This is where we get into the nuts and bolts of the company Quality measures. It’s long on theory, no modeling (again). But it’s important. You ability to properly incorporate these ideas into models will go a long way toward making the difference between models that implode when the market gets antsy, or are able to withstand shocks and live for another day. Next time, we’ll get back to modeling.


In Topics 4A and 4B, we introduced the topic of company Quality and considered how it relates to our task, identifying potentially attractive stocks. Quality comes together in profitability ratios, which on Portfolio123 include Return on Equity, Return on Investment and Return on Assets. One way or another, all three tell us the most important things we need to know about a company, how profitable it is, and it does so by controlling for variations in the key business input, capital. Through the Dupont Framework, we learned that Return on Equity (ROE), the bottom-line-type metric, is a function of Margin, Turnover and Leverage, the three measures that will be discussed in this Topic.

A Fly in the Ointment

Life would be easy if we could simply sort companies by ROE and pick from the top. Sadly, that’s not an option. Because our investing success depends on the future, knowledge of past ROE cannot be relied upon. The best we can do with it is hope that such knowledge can be at least somewhat extrapolated into the unknown. We saw in Topic 4A that we probably can do more in this regard than with many other metrics. But when we’re narrowing down to a small, investable, number of stocks, there has to be a limit even here to our willingness to pay it forward.

The DuPont framework helps by focusing on less grandiose measurers which may be more graspable, an important step considering that we need to use the past as a basis for making inferences about what’s to come. But even that’s imperfect. We need to be on guard for indications of possible deterioration in any of these three DuPont components.

Now, here’s where it really gets challenging. No single datapoint is worth much on its own, unless or until we have evidence that said datapoint is “normal;” i.e. not being influenced by temporary oddities. And the farther we go down the income statement toward the bottom line, the more opportunities there are for distortions to assert themselves. And I’m not talking about managerial manipulation – that will be discussed separately under Earnings Quality. I’m talking about a countless variety of whacky things that can throw any set of numbers off for any number of completely legal and ethical reasons, ranging from business conditions to company strategic shifts and a whole lot more.

This is why it is not necessary and may even be highly undesirable to model based on the full-blown DuPont equation: ROE = (Net Income / Sales) * (Sales / Assets) * (Assets / Equity). Other finer ratios relating to margin, turnover and leverage may – and often are – more effective when opening windows to probable futures. So as we discuss these ratios, I’ll focus on what’s useful regardless of whether it strictly conforms to the DuPont equation.


Let’s start with a quick review of the various margins available for use in Portfolio123 as pre-set factors:

  • Gross Margin: This is Gross Profit divided by Sales. Gross Profit is sales minus “cost of goods sold” (COGS) or “cost of sales, “cost of revenue” or any other label that sounds similar. The costs with which we’re concerned are the ones most directly associated with what is needed to generate the revenue. A good example is what a retailer spends for merchandise. This is a variable cost; it moves up and down along with revenue. Many COGS items are variable but often, the line between variable and fixed gets blurry. Crew and fuel expense, for example, come under COGS for an airline and vary in a broad sort of way with Revenues. More flights mean more revenue and more COGS. But within a specific flight, These COGS items are fixed without regard to how many seats are filled and how much each passenger paid. But whether fixed or variable, it should be clearly understood that these costs are the ones most closely and directly associated with Revenue.
  • Operating Margin:If you are interested in details, you must ALWAYS check the glossary of whatever source you consult since variations in how this can be computed – among companies, among data vendors, among web sites, among analysts, etc. – can seem countless. In a very broad sense, this is Operating Income divided by Sales and Operating Income is Gross Income minus overhead-type expenses; i.e., ordinary and necessary business expenses that are not really tied to revenue. “Selling, General and Administrative” (SG&A) expenses are the most commonly accepted major component here, but be aware that labels can vary considerably. For those interested in EBITDA margins (something that has become popular in the current generation), that, too, is available in Portfolio123. There is, however, a strong case for using Operating Margin, which reflects the subtraction of Depreciation. Those inclined to dismiss Depreciation do so because it’s a non-cash expense. Strictly speaking, that’s true. In reality, though, companies do make cash capital expenditures and CFOs typically encourage analysts to regard Depreciation as a reasonable proxy for ordinary and necessary capex, at least “maintenance capex.”
  • Pretax Margin: This is easy. It’s Pretax Income divided by Sales.
  • Net Margin: This, too, is easy. It’s Net Income divided by Sales.

Of the four main choices, the easy ones, Pretax Margin and Net Margin (the one that is part of the purest form of the formal DuPont formula), are least likely to be useful, especially Net Margin.That’s because they are the ones that bear the biggest brunt of all the oddities that can make an individual results unrepresentative of the company’s underlying fundamental characteristics.

The most relevant margins are those that sit highest in the income statement closest to Sales, the figure that’s least impacted by oddities. This doesn’t mean we’re blasé about possible non-representative aspects of reported sales. Entry into new markets can temporarily boost sales just as well as net income. Recessions hit Sales too. But whatever distortions do exist at the top are likely to be less severe than what we see on the bottom, since the sales precent movements are not magnified one way or the other by intervening items that don’t vary at all or vary to much lesser degrees.

Table 1 illustrates how the magnification of oddities can play out.

Table 1

Yr 1 Yr 2 % Ch Yr 1-2 Yr 3 % Ch Yr 2-3
Sales 100.00 120.00 +20.0% 138.00 +15.0%
– Cost Sales 40.00 46.00 52.00
= Gross Profit 60.00 74.00 +23.3% 86.00 +16.2%
-SGA 5.00 5.20 5.60
=Oper Prof 55.00 68.80 +25.1% 80.40 +16.9%
-Interest Exp. 35.00 35.00 35.00
-Other Stuff 12.00 3.00 3.00
=Pretax 8.00 30.80 42.40
-Taxes 2.80 10.80 14.80
=Net Income 5.20 20.00 +284% 27.60 +38.0%

Don’t for a minute doubt the realism of this example. Those who are accustomed to looking at 10-Ks and 10-Qs know that if anything, this is a very restrained example of the sort of erratic trends one sees pretty much as matter of course.

Table 2 shows the margins.

Table 2

Yr 1 Yr 2 Yr 3
Gross Margin 60.0% 61.7% 62.3%
Operating Margin 55.0% 57.3% 58.3%
Pretax Margin 8.0% 25.7% 30.7%
Net Margin 5.2% 16.7% 20.0%

So which margins do you think are most likely to be useful in helping you identify stocks with a view toward achieving satisfying live (out of sample, future) performance?

The answer lies in remembering why we care about margins in the first place. We are not interested per se in seeing how the company did. We are not the reporting or performance police. We don’t care about pointing fingers and saying “You were great” or “You stunk up the place.” This may not always be an easy mindset to control. After the big research and accounting scandals of the late 1990s and early 2000s, many, especially those with soapboxes, assumed the role of moral watchdogs. We need that sort of thing. But we don’t want to have such considerations enter into our modeling except the extent we knowingly and thoughtfully put them there (as we’ll do when we get to “earnings quality”).

What we want now is to get handle on how ROE might perform in the future. So the huge Year 1-2 boost in net margin is irrelevant. We cannot conclude from it that ROE is rallying like a space ship that just basted off. The business itself, the underlying core of the company, the part that over the long term is most likely to prove sustainable when all is said and done, is stable with a very gentle upward slope. But that slope is so gentle, we may be better off not counting on it and treating the company as one that is just plain stable (especially since we understand that the mega trend in ROE is a slow reversion to the business norm).

Trends in gross and operating margins support such a conclusion. Trends in pretax and net margins do not and can actually lead us to incorrect conclusions, even though they are accurate figures. The moral of the story: Reporting accuracy, while necessary, is not even close to being a sufficient condition for usefulness to us.

This raises a fascinating question of what it is we’re trying to analyze: a business or a company. They are not necessarily the same thing. The “business” is exactly what you’re likely to think it is, the goods or services the company sells, the revenues it gets from doing it, and the associated costs (direct and indirect or overhead). The “company” is all that plus “corporate stuff.” The latter includes at the very least its financing/balance-sheet strategy as well as other “strategic” things (non-core investments, and acquisition policy, restructuring efforts, and so forth).

There’s nothing wrong with doing analysis and creating strategies that aim at the “company.” Many do it. In my opinion, through, since we’re dealing with the unknown future, we better serve ourselves by decomposing the total and increasing visibility on “business” and “corporate stuff” separately. Decomposing enhances our opportunities to evaluate the potential past-to-future persistence component by component (and this, actually, mimics the rationale for using Dupont factors at all instead of just sticking with ROE).

Relative to turnover and leverage, the other DuPont components, margin enjoys no inherent superiority. A company that gets a 20% ROE through a 0.5% margin and massive turnover is no better or worse than another that gets the same 20% ROE but accomplishes it through the combination of an 18.5% margin and slow turnover. That’s part of the theory of Value.

But remember, stock price is not based on Value alone. Its Value plus Noise. When we consider the latter, we would be well within our rights to contemplate the possibility that margin is a much better understood more easily grasped concept than turnover. The latter is easy to define but not always easy to visualize, especially the concept of asset turnover (it’s easier to create examples that drive home the meanings of receivables or inventory turnover). So if your testing efforts suggest to you that your models are more responsive to margin than turnover, you could legitimately accept that conclusion without being vulnerable to accusations of data mining.


Turnover represents the pace at which a company is able to convert its assets into sales. Yeah, right! Now you see what I meant when I suggested above that Turnover is a more esoteric harder-to-grasp notion than margin and why the theory of Noise would allow you to omit it from your ROE-related modeling.

Actually, though, the fact that many do find it harder to grasp is something that might lead us to less-crowded trades with alpha-generating potential. We could do that, for example, by looking for stocks whose ROE trends and prospects are more favorable than many who focus only on margins might realize. Such strategies would likely require more patience since we wouldn’t expect the stocks to deliver until turnover can manifest in ROE trends that manifest in earnings trends, and hopefully in positive estimate revisions or surprises or moving average crossovers or gap ups etc. But this is a potentially fertile area for the creative among us.

The best way to visualize turnover in graspable terms is to drill down from asset turnover to receivables or inventory turnover.

  • Receivables Turnover (or Accounts receivable Turnover):This is Revenue divided by Accounts Receivable. It tells us how quickly the company is collecting from customers who buy and take delivery immediately and have to repay when billed, within 30 days or some other term agreed to by the parties. Great receivables turnover occurs when customers take fewer days to make their payments. Turnover slows as they wait as close as feasible to the 30-day deadline. Slower turnover is when they pay late. Bad turnover, turnover problems, occur when the bills stay outstanding for ever and ever often until the company realizes it can’t get blood out of a stone and writes off the receivable as uncollectable. This is a normal, albeit undesirable, part of business so we won’t freak out every time it happens (in fact, it’s so routine, many companies estimate bad receivables ahead of time and record them on their balance sheets). But it can’t be allowed to run out of control. Companies that make a habit of boosting sales by selling to too many customers that won’t be able to pay up can wind up in bankruptcy themselves (see, e.g. Lucent, for which badly deteriorating receivables turnover was the red flag that signaled eventual disaster).
  • Inventory Turnover:This is Cost of Goods Sold divided by Inventory. It measures how quickly goods move off the shelves (and into the possession of what hopefully will be a paying customer). It’s a very important metric but one that often attracts popular attention after it’s too late; i.e. after the news is out and the stock got pounded. It’s not as if the media and the masses suddenly come to understand what inventory turnover is. What they do understand, however, is that if goods sit too long on the shelves, management will have to take markdowns, possibly big markdowns (very big markdowns or even writeoffs if the situation is sufficiently dire to cause the company to sell to clearance outlets or to scrap goods) which will hurt margins which will hurt ROE which hurts profits which leads to negative earnings surprises or downward guidance and revisions – and it’s the latter that everyone understands. Phew, that’s a mouthful. But it’s real and sometimes very visible – especially for retailers that misjudged the market and wind up having to take their public lumps (especially after what had been hoped would be a peak selling season). For us, it’s another opportunity for creative minds. If you can find ways to get ahead of the crowd in this area, you’re likely to wind up with alpha.

And now, for the big item:

  • Asset Turnover:This is Sales divided by Total Assets. It tells us, aw heck, I really can’t make it any more concrete beyond the formula. I suppose, if it helps, you can think of it as how many times in a 12 month period a company sells its entire asset base. I’m not sure how much, if at all, that helps. Hopefully, though, your consideration of Receivables and Inventory turnover convinces you that this whole notion is pretty important, even if you can’t readily explain it to a 12-year old. And as strategists, we really need to recognize it because there are many service- and/or cash-transaction-oriented businesses for which receivables and inventories are of little consequence, meaning that Asset Turnover is the only solid chance we have to tune into this important measure of fundamental health. As with the other turnover ratios, you can pass on it and count on market Noise to justify your use of margin only. But for those really willing to work with Asset Turnover, such efforts could turn out well worthwhile.


This involves all things relating to financial strength. Strictly speaking, in DuPont terminology this is Assets divided by Equity. But consider this:

  • Leverage = Assets / Equity
  • Assets = Debt + Other Liabilities + Equity
  • Leverage = (Debt + Other Liabilities + Equity) / Equity

Presumably, you can see where this is going. The main driver of the numerator of this equation, in terms of analyzing company fundamental characteristics, is Debt. And if we step outside the strict DuPont framework and switch to a broader level of analysis, we’re looking just at Debt, and more particularly whether there is too much Debt.

Debt is an extremely important barometer of return-risk choices. Companies that borrow heavily stand to prosper more handsomely if things go well. But if there are problems, they can fall faster and harder. It’s the same way it is for investors who use margin. They do better in good times, but can more easily wipe out if Mr. Market acts ornery.

We don’t have to consider leverage, debt, etc. in every model we build. We can, if we wish, simply choose to swing for the fences and maximize potential return. And to the extent we do wish to factor risk into what we do, there are indirect ways to get at it. Large-cap orientation, for example, or focus on certain so-called “defensive” sectors (the “Population Growth” Smart Alpha theme) can point us toward companies that are most likely to have little debt and/or have cash flow streams that can comfortably support the debt they have, even if it’s a lot. We can also try statistical measures such to identify stocks whose past performance has been consistent with what we expect from moderate-risk companies.

When, however, we do choose to explicitly address leverage, the question we want to address, for our future-oriented approach, is not necessarily a strict DuPont-type Assets/Equity relationship but the one that answers the question: How comfortably can the company live with whatever debt it has.

The starting point for this kind of inquiry is typically one of the Portolio123 debt ratios, DbtLT2Ast, DbtLT2Cap, DbtLT2Eq, DbtTot2Ast, DbtTot2Cap or DbtTot2Eq. Don’t lose sleep over whether Dbt is compared to Ast (assets), Cap (capital) or Eq (equity). There are no analytically significant differences.

The issue of DbtLT (long-term debt) versus DbtTot (total debt) is a more interesting one.

Back when I started in the business in 19-none-of-your-business, the answer was clear. Long-term debt was debt due more than a year into the future. This was part of the company’s permanent capital and hence, ripe fodder for basic fundamental analysis. Short-term debt, debt due within the next 12 months (the extra borrowings that make up the difference between DbtTot and DbtLT) was not considered part of the permanent capital. Stereotypically, this was temporary trade credit, such as what a retailer would use to finance the purchase of inventories. As the goods were sold, the debt (often a revolving line of credit) would be paid down.

But as corporate finance departments embraced information technology and as interest rates embarked on a multi-decade slide (making companies reluctant to lock themselves into borrowings at rates higher than what they figured they could get a year or so down the road), they became increasingly willing to use debt nominally-designated as short term as part of what they, internally at least, considered permanent capital. And as banks got increasingly creative in finding ways to make themselves more attractive to corporate customers at the same time they, too, got better at I.T., they often became less finicky about money used to finance working capital being paid back and taken off the books in less than a year.

Nowadays, the distinctions between DbtLT and DbtTot (DbtLT plus DbtST) have not vanished. But they are not quite as hard and fast as they once were. That complicates our lives a bit. Since we can no longer make an absolute good-for-all-times statement about whether DbtLT or DbtTot is better for our purposes, but we know DbtSomeSort is important. This is an area in which one’s choices could legitimately be guided by experimentation and testing.

Once you’ve made a choice, let’s say DbtLT2Eq, what do you do with it?

The usual default answer is “lower is better.” All else being equal, that’s true. But all else is rarely ever equal. Higher debt ratios can be justified, or even preferable depending where in the return-risk continuum you want to wind up, if the company’s business is one that is better able to handle debt. This is usually the case with businesses characterized by more stable, often less cyclical, cash flows. This is why the market tends to be tolerant of heavily leveraged utilities, consumer non-durables and large media companies.

For this reason, you will often want to use debt ratios that are compared to some sort of business (sector, subsector, industry or subindustry) benchmark either through Ind or the FRank function. A debt ratio alone cannot tell you if debt is excessive. A debt ratio that is high relative to others in the same line of business (and presumably with similar cash-flow characteristics) is a step in that direction.

This doesn’t mean you can’t use a debt ratio on it’s own. That’s a perfectly fine thing to do if you are willing to (or may even want to) skew your results toward companies that go out of their way to put themselves at the conservative end of the reward-risk paradigm regardless of business exposure.

Another decision involves whether we even want to bother with debt ratios per se. The point of doing so is based on a presumption that the greater the debt as a percent of the capital structure, the greater the interest burden will be. There’s nothing wrong with that presumption. Ultimately, though, it is just a presumption. We have an alternative. We could measure the interest burden more directly. Interest Coverage (IntCov) is a Portfolio123 pre-set factor that allows us to do this. It’s calculated as Operating Income After Depreciation divided by Interest expense. A variation with which you might want to experiment is Cash From Operations divided by Interest Expense (a ratio that measures surplus availability since interest is subtracted before we get to cash from operations).

It’s tempting to completely abandon the capital ratios and go entirely with coverage especially considering that the ratios discussed above already adjust for unusual items. But there are other factors (exceptional business conditions) that could make any individual period unrepresentative. That argues in favor of continuing use of capital ratios, which tend to be more stable over time. (In this regard, bear in mind that debt ratios tend to be management strategic choices. Don’t expect long-term debt to ever be repaid – unless or until management changes its financing strategy. When borrowings mature, they tend to be replaced by new financings).

Return on What?

Portfolio123 offers pre-set factors for three different returns:

Return on Equity (ROE):This is Income Available for Common divided by Equity.

Return on Investment (ROI):This is Income After Taxes divided by the total of Long-Term Debt and Stockholder’s Equity.

Return on Assets (ROA):This is Income After Taxes divided by Total Assets.

You can find additional variations on Wikipedia, Investopedia, or through any number of other sources. The details of any particular formula, a Portfolio123 formula or one you find elsewhere or create, are not likely to make or break any strategy you create. What’s most important is that you understand what each variation tries to accomplish. All are intellectually valid.

ROE is the granddaddy, the focus of the DuPont framework, the metric upon which we’ve been focusing so far. It is the penultimate measure of the quality of a “company.” ROI and ROA, on the other hand, vie for stature as the penultimate measure of the quality of a “business.”

Similar to what was discussed when we compared gross or operating margin versus pretax or net margin, a company is a business plus corporate stuff. Both are important. The business is what’s likely to be more persistent from past to future. But the shares that trade do not relate to the business; they attach to the company. Which is more important? Both of the above! It’s not necessary to throw every possible thing into every model. But as long as you understand what you’re getting, you can go either way or both ways. (Inadequate attention to the “all else” that makes higher return-on-something better than lesser return is the likely difference maker.)

A Post-Script on Risk

On various occasions, I’ve said that I regard fundamentals as a far better way to measure risk than statistical calculation based on historic price data. Hopefully, by now, you have a sense of why. ROE (and ROI and ROA) tends to be persistent. And the things we’re looking to accomplish when we model using Returns and DuPont components tend to relate to the stability of an acceptable trend.

That is what risk reduction is all about. Debt-heavy companies have higher fixed-cost burdens which lead to more volatile EPS and cash flow trends which create more opportunity for significant surprises and estimate revisions, and that causes sentiment to change often and sizably, and that is what makes for share price volatility. Simply put, Beta and Standard Deviations are laggingindicators of risk. Debt is a leadingindicator of risk. So too are other factors that impact the stability of the earnings trend. Since equity investments succeed or fail based upon the future, it stands to reason that we should prefer leading indicators of risk.

This makes for an opportunity to deviate a bit from the DuPont framework and consider balance-sheet liquidity. It doesn’t really impact ROE. But it is a huge indicator of financial risk. In fact, I might go so far as to say no company ever went bankrupt due to a high debt ratio. Liquidity crunches are what make management throw in the towel and tell the lawyers to head for court.

An example of how this works was my experience with Rite Aid (RAD), a stock I aggressively recommended in a newsletter at a price of about 1, which turned into something close to a ten-bagger. At the time I started pushing it, the consensus view was that the debt ratios were astronomical (and even unquantifiable given that equity was a big negative number) and that there was no chance the company would ever be able to repay a meaningful portion of it. What that analysis missed was liquidity. RAD had lots of it and was very much cash-flow positive. Of course they would not be able to repay the debt but we covered that – the default expectation calls for refinancing. Survival was tied to the willingness of creditors to roll over debt, or modify its terms if need be. And that’s the key: If creditors are willing to finance the operation, a company can flourish with any debt ratio, no matter how “bad” it seems. And whether or not they’ll do this depends on their assessment of the company’s business performance and prospects. RAD was seen as doing the right things operationally, so the liquidity continued to flow. I’ve seen plenty of companies with far “better” debt ratios go into Chapter 11 within hours after trade creditors expressed reluctance to continue to deal.

So when modeling for risk, don’t stop at debt ratios. Be aware of the basic pre-packaged liquidity ratios Current Ratio (current assets, mainly cash and cash equivalents, receivables and inventories minus current liabilities, mainly accounts payable and short-term debt) and Quick Ratio (this is similar to Current Ratio except that here, Inventories, the potentially hardest-to-monetize part of Currant Assets, are subtracted from Current Assets before it’s divided by Current Liabilities). These indicate the speed with which impermanent assets could be quickly monetized. If you see a deteriorating Quick Ratio, do not buy the stock for widows and orphans even if Beta is well below 1.00. (Note though that the some businesses have financial characteristics that de-emphasize the balance-sheet items on which these ratios depend. So be aware of industry comparisons.)

And, of course, don’t forget basic everyday business performance. As per the RAD example, this is what risk is all about.

Next Up

We’ve had a lot of theory. Next time, we’ll see how such ideas can play out in models. And prepare to manage expectations. We won’t be looking for super-alpha moon-shots. We’ll be aiming for increased probability of future (live or out-of-sample) performance and diminished risk.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s