Banner Default

By: DC Croson

Cronus at the closing table

In Hesiod, Cronus swallows each of his children as it is born, having been warned that one of them will overthrow him. Modern incumbents have a gentler procedure. They pay the child's investors first.

Consider Questcor, which in the early 2000s held a monopoly on ACTH drugs through its product Acthar, used for rare conditions such as infantile spasms. In 2013 it bought the US development rights to Synacthen, a synthetic rival then working its way toward the American market. Questcor did not develop it. The Federal Trade Commission later complained that the purchase had thwarted a nascent challenge to the Acthar monopoly.

Colleen Cunningham, Florian Ederer and Song Ma call this a killer acquisition: buying an innovator in order to stop its project. Their 2021 paper in the Journal of Political Economy follows more than 16,000 drug projects from over 4,600 originating firms, 1989 to 2010, and tracks each one through any change of ownership. Projects bought by a firm that already had a drug for the same disease working by the same mechanism were 23.4 percent less likely to show continued development than projects bought by firms without such an overlap. The authors' conservative estimate is that 5.3 to 7.4 percent of pharmaceutical acquisitions, roughly 46 to 63 deals a year, are killers.

Why a buyer would pay to bury

The logic is old; the evidence is new. Kenneth Arrow pointed out in 1962 that a monopolist gains less from an invention than a newcomer does, because the new product cannibalizes profits the monopolist already earns. Economists call this the replacement effect. Richard Gilbert and David Newbery added the other half in 1982: an incumbent has more to lose from a rival's success than the rival has to gain, so it will outbid the rival for the rights. Put the two together and the incumbent buys the project, then declines to develop it. The purchase price is an insurance premium against creative destruction. Entrants will sometimes deliberately create a company as bait for such an acquisition, knowing it’s easier (and cheaper) for the incumbent to buy them out than to compete with them indefinitely – effectively being a gadfly for profit. (This is an especially common story for serial entrepreneurs with deep skills in a particular technical area – then they reinvest the proceeds in their next venture.)

The data line up with the theory at every turn. Acquisitions are almost four times as likely when the buyer's products overlap the target's. The drop in development is concentrated where the buyer has the most to protect: markets with few competing drugs, and buyers whose own patents are far from expiry. And the terminations are swift. Overlapping projects are 20.9 percent more likely to show no development at all after the deal.

The most telling finding concerns antitrust review. Under the Hart-Scott-Rodino Act, small deals need no premerger filing with the FTC; in pharma, deals under an inflation-adjusted $200 million (in 2000 dollars) usually escape review. Overlapping acquisitions bunch just below that line, and non-overlapping ones do not. Deals just under the threshold went on to launch a product 1.8 percent of the time, against 9.1 percent for deals just over it. This $200 million cutoff covers just about everything that would be worthwhile to buy and abandon; above that threshold, if you’re going to pay that much buy it (and be subject to antitrust scrutiny), you might as well use it. It’s the small fry which are the bargains for the acquirors.

The authors even test whether buyers were really after the target's technology or its scientists (often called an "acqui-hire") yet found little sign of either: only 22 percent of the target's inventors ever worked for the acquirer, and acquired molecules generally did not reappear in the buyer's later projects. 

Let's not forget, however, that the prospect of being bought out, even to be buried, is part of what lures investors (and the serial entrepreneurs mentioned earlier) into science-based startups in the first place. It's a lot easier to cash the acquiror's check than to build the business and compete against them.

Software keeps the engine and scraps the car

A drug project is developed or shelved; there is little in between. Software (an obvious complement to sensing technologies) allows a finer cut. A buyer can incorporate the parts of a product it wants into its own and simply switch off the rest without cost or notice.

Axel Gautier and Joe Lamesch counted 175 acquisitions by Google, Amazon, Facebook, Apple and Microsoft between 2015 and 2017. In most cases the target's product was discontinued under its own brand; the younger the target, the likelier that end. Their reading is that the giants mostly buy functionality, technology, and talent to fold into their own ecosystems, so that acquisition serves as a cost-effective substitute for in-house R&D. 

They found only one deal in their sample that plausibly fit the pharmaceutical killer pattern, however.
Software's characteristic deal is the partial burial: the most-loved features survive but the product does not. Users of Sunrise (acquired by Microsoft) who wanted the unique features kept them, provided they moved their email to Outlook. Dark Sky's (acquired by Apple) forecasting engine lives on, but only for people who own Apple hardware, which continues to sell at a price premium; PhoneArena reported that the Android version had more than a million users when it was switched off by its acquiror, Google.

Which features survive, and why

Arrow's replacement effect does not disappear when a buyer integrates instead of killing. It moves down from the product to the feature. A buyer keeps a feature when it makes the buyer's own platform more valuable and drops it when its value lies in working with someone else's.

The cases sort neatly on that rule. Dark Sky's precipitation model made iPhones more attractive; its Android app made Android phones more attractive, and its API improved other companies' weather apps. Sunrise's calendar views improved Outlook; Sunrise's ability to ride on top of a rival's mail service removed a reason to switch to Outlook. What ends up on the cutting-room floor is, as a rule, whatever complemented a rival. Dark Sky went dark only on the far side of the fence.

There are a few interesting examples in the wearable sensor market, particularly Apple Watch and Fitbit. Apple’s purchase of Beddit, a Finnish maker of under-mattress sleep sensors, played out the Dark Sky script in slow motion. Beddit’s strip worked with any phone. After the acquisition, Apple gradually stripped out features such as cloud sync and Android support and discontinued the hardware in 2022. Sleep tracking reappeared on Apple’s own device: for years the Watch had depended entirely on third-party sleep apps, until Apple built native tracking into watchOS 7. Apple never said what it took from Beddit, and observers assumed its patents and engineers had been absorbed into Apple Watch development. What survived was sensing know-how that gives people a reason to own an iPhone and an Apple Watch – high-margin hardware items. What was dropped was the part that served Android owners and anyone unwilling to wear a watch to bed. 

Fitbit’s 2016 purchase of Pebble, the Kickstarter-funded smartwatch pioneer, was a partial burial from the first day. Fitbit acquired Pebble’s software, firmware, key staff and intellectual property, but not its hardware. Pebble’s next models, the Time 2 and Core, were canceled and Kickstarter backers refunded. Pebble’s operating system became the foundation of Fitbit OS. The rest of the ecosystem was switched off by June 2018, including the app store, the forum and voice recognition, and even text and email replies. Of the community’s watch apps, Fitbit carried over only three. Pebble’s founder had called the match natural, citing values the firms shared, among them working with both iPhone and Android. Cross-platform support survived. The independent developer community built around Pebble’s own store did not. When Fitbit later bought Coin (a payment provider) Fitbit apparently didn’t use any of the softwear at all: they took only certain assets and IP to add payments to its wearables and discontinued Coin’s wallet.

Buying a team and its best ideas is often cheaper than building them, and many users got the features folded, at no charge, into an app they already had. Who lost, though? Users of rival platforms, developers who built on an API, and the next founder deciding whether to enter.

Beware of buyers bearing features

Laocoön warned the Trojans, timeo Danaos et dona ferentes: he feared the Greeks even when they came bearing gifts. The pharmaceutical killer acquisition at least looks like what it is, since a promising drug simply stops appearing in the trial registries. Software's version arrives as a gift: your favorite feature, free, in an app you already own. The question for users, regulators, and the next generation of founders is what should be welcomed -- and what must be left outside the walls.

Sources