In the 1960s, there was a market vogue for conglomerates: sprawling, acquisitive businesses that mastered operations in one domain and then applied their playbook to others. At least, this is what people said about conglomerates when they were flogging the common, convertibles, debentures, tracking stocks etc. that these businesses naturally produced.¹

The bull case for conglomerates was that they were applying the new science of management to older companies, and that the more industries they were involved in, the more insulated they were from industry-specific downturns—an airline needs plane tickets to stay expensive and fuel and labor to stay cheap, but an airline that diversifies into oil, and acquires a discount retailer, is hedging those risks out. The moderate case, paraphrased in modern terms, is that they were a way to structure a levered long small-cap value trade: it was easier for a conglomerate to get massive leverage than for a trader to, and they wouldn't get margin calls if the value of their assets fluctuated, but there were still lots of cheap companies out there for them to buy. The maximally bearish case was that they were performing a boring math trick, by using their overpriced stock to buy slightly cheaper and slower-growing companies, which optically increased earnings per share—if you use stock trading at 50x earnings to acquire something for 35x earnings, your earnings go up. And the maximally bearish case was:

1. Sure, you can do that, but over time your P/E ratio comes down because more of your business is slower-growth, and the only way to keep the merger-driven EPS machine going is to keep buying cheaper companies, but also to keep buying bigger companies. So you start with a small electronics business, turn into a mid-sized defense contractor, and then end up being a sprawling meatpacking, insurance, and wire company that's trying to buy a declining steel company at a single-digit P/E. (That was LTV.)

2. Antitrust enforcement was very strict at this time, so companies couldn't do strategically sensible mergers where they buy suppliers, customers, or competitors. Instead they had to do things like diversify from telecom equipment into rental cars and Wonder Bread (that's ITT). But since the operational improvements are mostly fictional, it doesn't really matter that these businesses have nothing to do with each other.

That model of diversification has mostly died. A handful of companies still practice it, but there are basically three modern variants: PE firms acquire companies in multiple industries, but run them independently and generally intend to sell them separately—there are times when they plug companies together, whether as a roll-up or as a way to avoid booking a loss on a deal, but roll-ups are mostly horizontal integration andaren't usually driven by strategic considerations.² There are companies like Constellation Software, Roper and Danaher that have a distinct operating model, and part of their growth model is that they can hire talented middle-managers and offer them a credible path to promotion—if you're choosing between two companies with 5% organic revenue growth, the one that adds on 10% acquisition-driven growth is more likely to promote you to a position where you have more responsibility. Ironically, this means that M&A-driven growth has switched from being justified by the science of management to treating it as something closer to an art, where you want a lot of young talents to show off their skills before you decide which of them deserves a bigger canvas.³

The corporate finance puzzle around diversification is that there are two ways to get it: a company can buy it for its shareholders by paying a 30% premium to buy out some other public company. Or shareholders can pay a few basis points to diversify into that company (if they want, or to hold on to their shares if they don't). Narrowly-focused companies increase investor choice, since that gives them a way to bet on specific industries, or even specific links in the supply chain.

The argument for within-company diversification as a hedge against industry fluctuations is also a weak one that shareholders can handle themselves. If companies are well-capitalized enough to ride out a downturn, they don't need to hedge by owning another business they don't understand as well. If shareholders want to juice returns, they can lever up themselves instead.

Strategic mergers—buying a direct competitor, a supplier, or a customer—are diversification in one sense, but it's a weak one. If Apple buys the supplier of a critical component, they're reducing the risk that Samsung will outbid them for that supplier's output in the future, or that the supplier will go under or, for whatever reason, stop making the product. But this also means doubling down on end demand for Apple's products, since now more of the cost of goods sold and opex will flow through Apple's own P&L instead, and, in the event of a steep drop in demand for iPhones, Apple's fixed costs will be higher. And this usually isn't pitched as diversification—it's usually pitched as a way to align the target company's R&D and production even more closely with Apple's product plans.

One of the historic reasons that companies would diversify in a non-strategic way was that they were buying job security by reducing single-industry dependence and by making the job of managing that company trickier (if you'd somehow decided to run an activist campaign against Berkshire in the early 80s, a good argument against you would be "where else can you find someone who knows a lot about textiles, insurance, newspapers, and chocolate?"). But that argument, at least, has weakened, because shareholder activists see this kind of diversification and smell blood in the water. A CEO who acts like their job is at risk because of future earnings disappointments is one who ironically put their job at nearer-term risk from an activist who can achieve an immediate pop in valuation by unwinding a bad deal.

So when companies diversify today, they usually do make a strategic argument. Investors are happy to own shares of a business that can't keep expanding, so long as the cash flow it earns is devoted to making the share count shrink instead. Dumb deals still happen, because the incentives are there, but if there's alpha in killing those deals and alpha in avoiding them, they'll keep getting rarer.

We’ve covered diversification in its many guises over time in The Diff:

• Here’s a look at Cerebras, which was basically an AI side bet ($).

• How Danaher got diversification right ($).

• And Teledyne’s approach ($).

• In equities, cheaper trading means that it’s cheaper for indexers to diversify but also less risky for stock pickers to concentrate ($).

Join the discussion!

1  "Tracking stocks" were a popular tool in the 90s when the market applied a premium to more dot-com-flavored businesses that were buried in larger conglomerates, and they still exist in a few cases today. They're convenient in a sense, but it's annoying to read one 10-K that references three different tickers and having to keep jumping around, so they tend to work best when investors are excited about themes rather than interested in the present value of future cash flows.

2  PE firms like to be able to say that their purchases turn out well on an individual basis, because this helps them negotiate an acquisition—fewer people will leave a PE-owned company if they think of PE as buying growth rather than harvesting cash flows. And PE firms really don't want to be seen as financial engineers taking advantage of limited liability to make a bunch of independent bets—if they ran every portfolio company so its value went up or down by 300%, they'd capture that 300% upside but only the first 100% of the downside.

3  A unique part of Constellation’s incentive structure is that top-performing managers can run their own portfolios of businesses — like mini-CEOs. The model creates new business unit managers constantly. From Mark Leonard’s 2017 shareholder letter: “This is the point at which our Operating Group Managers or Portfolio Managers can provide coaching. If a large BU is not generating the organic growth that we think it should, the BU manager needs to be asked why employees and customers wouldn't be better served by splitting the BU into smaller units. Our favourite outcome in this sort of situation is that the original BU Manager runs a large piece of the original BU and spins off a new BU run by one of his/her proteges. Ideally, he/she has been grooming a promising functional manager who’ll be enthusiastic about running and growing a tightly focused, customer-centric BU.” It is rare for other conglomerates to split BUs into smaller units often, but the model helps motivate junior folks at Constellation to perform and keeps management hygiene very high: “With a clean sheet of paper, the leader only takes those he needs. They set up in an open office with good communication and no overheads. They cover for each other. They leave all the bureaucracy and the crap behind”.