How the D20 Index Methodology Handles Mergers and Delistings

An index that handles every acquisition, merger, and delisting as a one-off decision isn't really a methodology. It's a series of judgment calls dressed up as one, and over enough years the inconsistency between those calls starts to matter more than any single decision did on its own. That's true of any index built around a moving industry, but it matters more here because autonomous vehicles are unusually prone to exactly the kind of corporate reshuffling a rulebook has to anticipate.

The alternative is publishing a documented rule for what happens when a constituent stops existing in the form the index originally tracked it in. Get acquired, and there's a rule for whether the acquirer replaces you or a different company gets added instead. Merge with another constituent, and there's a rule for how that collapses two slots into one rather than leaving an empty seat. Get renamed after a restructuring, and there's a rule for whether that counts as the same constituent under a new name or a fresh addition entirely. Delist, and there's a rule for removal rather than a debate that plays out differently depending on who happens to be paying attention that quarter.

Some of the hardest calls don't fit neatly into acquisition, merger, rename, or delisting at all. Uber didn't merge with anyone or get delisted when it sold its Advanced Technologies Group to Aurora in 2020; Uber itself kept trading, kept operating, and kept doing everything else it had always done. What changed was narrower: the specific autonomous-vehicle business that had earned Uber a place in the index stopped existing inside Uber at all. A rule built only around whole-company events, an acquisition of the company, a merger of the company, a delisting of the company, has nothing to say about a parent company divesting the one division that mattered to the index while remaining perfectly healthy otherwise. That case needs its own explicit rule: track the business, not just the corporate shell that happens to house it.

A different edge case shows up when a company pauses operations under regulatory pressure without formally shutting anything down. Cruise's nationwide pause following its 2023 California permit suspension is a clean example: for a stretch of time, the company hadn't shut down, hadn't been acquired, and hadn't delisted, yet it also wasn't running the driverless service that had qualified it for inclusion in the first place. A methodology has to decide in advance whether a pause of that kind triggers removal immediately, after some defined grace period, or only once it becomes clear the pause is functioning as a slow-motion shutdown rather than a genuine pause. Deciding that question in the moment, with a specific company's fate hanging on the answer, is exactly the kind of reactive judgment call a published rule is supposed to prevent.

What that documentation actually buys is auditability. Anyone can look at a constituent list and see who's currently included. Far fewer indices let an outside reader check whether the reconstitution decisions that produced that list were applied consistently, or whether the index quietly bent its own rules whenever a particular removal or addition proved inconvenient. Publishing the rule alongside the outcome is what makes that check possible at all, and it costs the index something in flexibility: a documented rule is harder to bend quietly than an undocumented habit would be, which is rather the point.

The D20 index has been through its share of acquisitions, mergers, and delistings over the years it has run, and the ten-year retrospective documents both the specific decisions and the rules that produced them. Some calls were closer than others.