In partnership with

In theory, an investor is allocating capital among all potential investments, which means you need some way to compare very different businesses. An airline is not a SaaS stock, but a dollar is a dollar and, implicitly, the market price of both stocks is the point at which the marginal investor is indifferent between investing in those two companies.

One good starting point for characterizing companies is the "DuPont formula," a simple tool for thinking about companies across different sectors. The approach is to look at a company's returns as a product: if you multiply a company's profit margin by its ratio of sales to assets, and then multiply that by its ratio of assets to equity, you’ll get a return on equity.

This is a tautology, like all accounting identities. And also like those identities, it's illuminating, because it breaks the general problem of measuring business performance down into a series of sub-problems that are more tractable. It also helps illustrate the difference between two kinds of businesses. Louis Vuiton and Costco, for example, have similar returns on equity (30% for LVMH, 28% for Costco). But they get there in completely different ways: Costco's sales are 3.6x assets, and LVMH's are 0.6x—LVMH just offsets its higher asset requirement for a given level of sales with much higher margins.

The last factor, the ratio of assets to equity (i.e. leverage), shows up because some businesses are entirely too boring without it. Buying a rental property without leverage might produce a return of, say, 5%, with gradual and unsteady increases in real rents over time. That business gets more interesting with more borrowing, allowing the investor to up their expected rate of return on equity from something comparable to t-bills to something closer to equities, or PE, or even higher than that.

In practice, it seems that companies choose whatever leverage policy pushes them in the direction of hitting the overall market's average return on equity, at least if they're a business bounded by some asset requirements. So software companies tend to hoard lots of cash, and their capital efficiency looks very different depending on whether or not that cash counts in the asset base. Real estate companies naturally lever up, which is partly a convenience for investors since it makes their beta closer to 1. Airlines tend to run with leverage, too, since they're geographically-bound businesses whose main asset has a global market.¹

So you can break the high-level problem of increasing return on equity down into three separate levers:

1. Increase asset efficiency

2. Raise margins

3. Lever up

And these three can be broken down further, albeit with more complexity. As the Costco/LVMH model shows, there's a tradeoff between offering a wide selection of slow-moving, high-turnover goods and offering a narrow selection of products that are certain to leave the shelves before suppliers even get paid. We can be a bit more general on the leverage side: one trivial way to increase leverage is to reduce asset efficiency—a company that switches from leasing its office to owning it will have upped its leverage at the cost of growing the asset base faster than revenue.

This model starts to get trickier when we look at cases where there's heterogeneity between the different products a company sells. A fancy restaurant, for example, might have a contribution margin of, say, 20% on food (i.e. if they seat one more diner at an otherwise-empty seat, and that diner orders a $50 meal, $10 gets added to the company's pretax profit after accounting for ingredients and labor). But they might also have a long wine list, with a 70% contribution margin. The ingredients and labor have a short shelf-life; you can barely move some of the labor burden around within a shift by doing some prep work, but ultimately run into inflexible barriers to doing more than that. The ingredients will also tend to go bad in not too long. But if it's a long wine list, the average bottle will stick around for quite a while.² This combination creates an incentive to treat the food as a complement to the wine, since pushing wine sales up simultaneously increases asset turnover and margins. So a recommended wine pairing is the purest synthesis of the culinary arts and the imperative to maximize return on equity.

That general setup describes the steady state in high-quality growth businesses. They'll often have one element that's resistant to growth through pure scale, but there will be a complementary part of the business that can absorb arbitrary amounts of capital. Turning money into software is not a direct process (ask any non-tech Fortune 500, any government agency, or, for that matter, a small business that tries to outsource building a website). Money can, however, turn servers into uptime, product performance, and model training in a more linear way.³

The DuPont formula is a very good way to get a high-level look at what makes a business as profitable as it is, and it's a good way to start comparing performance between peer companies and to understand what drives different industries' economics. The better-quality the company, though, the more likely its model is to rely on a complicated interplay between these different pieces. Feedback loops and unit economics drive the deepest analysis of what companies do and are capable of doing, while aggregate analysis like the DuPont formula tells you what the result was.

Worth the buzz: Amazon, Saas, and 12.1% Returns

Investors thrive on WebStreet despite market turbulence. Here's how: They connect accredited investors with expert portfolio managers specializing in Micro-Saas, Content and Amazon FBA, delivering passive income. Last year Webstreet Funds yielded a robust 12.1% cash yield, and Round 6 is underway. Their approach is proven but their offerings are limited. Join WebStreet now and find out what investors are buzzing about.

Read More in The Diff

The Diff has previously covered the importance of returns on capital, and the many ways to think about it. 

• Buying stock in a high return on equity company can be viewed as purchasing an option on future reinvestment.

• The Liberty Media empire is a good example of a cluster of different related companies with variable DuPont Formula results.

• Many great companies have a high-return but hard-to-scale business adjacent to a lower return but indefinitely scalable one.

• Aircraft leasing ($) is a great example of how balance sheets shift when there’s a slice of returns that’s predictable and can be shifted around.

1. They have another reason to use leverage, though: a young airline's margins will expand as it grows, because it's amortizing some fixed costs over a larger business and can provide cheaper redundancy. High growth also means that more employees with seniority-based pay scales are in the early, cheap part of their career. Which should imply that larger airlines would delever, because they can't outgrow their existing employee base. But literal leverage gives these companies a figurative kind of leverage: they're in a better negotiating position if they can honestly tell unions that there's a wage scale at which their credit will be impaired and the business will potentially enter a death spiral. It is, from management's perspective, better for overall costs to borrow money and buy back stock. Weirdly enough, since this reduces the marginal cost of flights, it actually represents redistribution from unions to travelers, with the company as an intermediary taking a cut.

2. They actually segment their inventory a bit, since some will be sold by the glass and some by the bottle. Lower margins are tolerable on high-volume products because the inventory carrying cost is lower.

3. And just to add the last big performance factor in: the next two months of an engineer's code contributions do not constitute collateral for a loan; a GPU of equivalent cost, however, can be financed. The GPU-rich companies are not cash-constrained at the moment, but wider insourcing of AI or sufficient growth in AI capex could eventually change that.

Join the discussion!