What kind of company is Amazon, and how many locations does it want?

We’re now into round 2 of the great Amazon HQ2 extravaganza, and as with the initial announcement much digital ink has been spilled to analyze the meaning of the winnowing of the list to a mere 20 cities. The Brookings Institution’s Jenny Schuetz has a very smart breakdown of the relative housing markets of the different contenders, splitting them into high cost places, recently gentrifying markets, those with abundant housing, and cities with older housing stock.

In our view, most of what’s been written about the Amazon HQ2 contest misses two important points.  First, Amazon’s decision is likely to be dictated by internal business strategy factors, especially what kinds of technologies and markets it wants to develop, and its choice of cities will be dictated by whether a city’s specializations match the particular functions Amazon chooses to move away from Seattle.  Second, Amazon is very likely to chose multiple locations (HQ2, HQ3, HQ4) to tap different city specializations, to minimize its negative effect on housing markets, and to continue to its negotiating leverage over “winners.”

Specialization matters

First, no one seems to be paying much attention to what business functions Amazon will place in its second headquarters. The assumption seems to be that HQ2 will be a functional clone of the Seattle mothership. That’s seldom, if ever, the way that corporations partition activities among different sites. Opening a second major headquarters location is a non-trivial event in Amazon’s business strategy:  which functions get spun off to a new location, and which are left in Seattle will make a big difference to the company’s future.

Simply duplicating all of Amazon’s Seattle operations in another city would make almost no business sense. Coordinating activities and managing rivalries between otherwise similar centers would invite an internal them v. us dynamic that could outweigh any advantages of a second location. Far more likely that Amazon will spin off discrete functions to its second location. Seattle will continue to do some (probably most) things, but HQ2 will develop its own strengths and specializations.

Viewing HQ2 as a clone leads most observers to treat cities as a kind of undifferentiated commodity, measured solely by a common set of generic attributes (their housing stock, workforce with a college degree, or tax break packages).  But if Amazon is really going to establish a second location with a distinct set of specializations, then the choice of a second city is much more likely to hinge on each city’s strengths as a base for that kind of activity. Our study of the high technology industries in the United States published by Brookings shows that the range of cities generally regarded as tech hubs (Austin, Boston, San Jose, etc), each exhibit distinctive technological specializations, and what’s high tech in one city is not the same as what’s high tech in another.

So the big question is this:  What kind of company is Amazon going forward? Over the next decade or two, what parts of the Amazon enterprise does the company want to grow that it can best nurture in a location that isn’t Seattle?

The answer to that question is complicated by the fact that Amazon is innovating at the periphery of several different industries:  retailing, logistics, software, and media/entertainment, to name just four.  Depending on which of those functions one wants to emphasize, or spin off to a non-Seattle location, one would focus on a very different subset of cities.

For example, if Amazon sees itself increasingly becoming a logistics company, competing with the likes of Fedex and UPS, then it might want to locate HQ2 in a mid-continent location that could become a national hub. (Amazon already operates its own delivery vehicles in many markets, and has chartered aircraft to expedite shipping). Indianapolis, Columbus, and to some extent Chicago, offer similar locational advantages to places like Memphis (Fedex’s hub) and Louisville (UPS).

Or if Amazon views itself as a software company, developing more robotics (like drone delivery), or further refining its techniques for marketing, then perhaps it will want to go some place like Pittsburgh.

But Amazon may view itself as a rival to established media and entertainment companies, in which case Los Angeles (movies, TV), New York (print, advertising), or Washington (news) may make the most sense.

Amazon may also be a telecommunications backbone company, building on its AWS (Amazon Web Services) division.  In that case, it might make sense to have a headquarters in Washington DC (strong government market, close to regulators) or Philadelphia (which is already home to Comcast).

The biggest factor at work here may be the kind of specialized labor force that Amazon could tap in each of these markets. The critical decision point may not be the simple availability of educated or creative workers, but rather the availability of just the kind of specialized workers Amazon wants. For example, Amazon reportedly told Detroit that it didn’t make the list of finalists because it didn’t have the kind of talent pool the company wanted. Richard Florida harrumphed that Amazon foolishly excluded Detroit from its list of 20 finalists, despite the fact that the region had hundreds of thousands of creative class workers.

If Amazon were looking to innovate in the automobile business, he’d have a point, but it may not be just general knowledge workers the company is seeking, but specialized talent in areas relevant to its expansion plans.

A Little History

It’s worth reflecting for a moment on the history of Amazon. Why is it in Seattle in the first place? Jeff Bezos came up with the idea selling books on-line in 1994, while living in New York. But rather than start his company there, he packed his bags and moved to the Pacific Northwest. According to a 1996 interview in Fortune, the choice was quite deliberate:

He chose the Seattle area because of its proximity to both high-tech talent and a major book distributor, Ingram’s warehouse in Roseburg, Oregon.

The next big thing is a bookstore,” Fortune, December 9, 1996

Most startups are established and grow in the places where their founders happen to be living. Few people consciously pick up and move to an entirely different community and start their business based on labor market and supply chain analyses. The fact that Bezos did so more than two decades ago, when his company was just a partly formed idea, is a good indication that it will be just as rational, now that its a huge and growing corporation.

Why not HQ2, HQ3, HQ4?

If specializations are important–and we believe they are–a critical question is why Amazon ought to create just one secondary headquarters. Part of the argument would have to be that there must be significant synergies between different functions that the company would reap a business advantage from having all those functions in a single location. In an important sense, however, the company has already crossed the Rubicon on that issue, by deciding on a major expansion outside Seattle. What that decision says, in effect, is:  “We don’t have a good business reason to expand all these functions in close proximity to our core.”

Once the company has committed to put some functions in HQ2, it may find it simpler and far more productive to establish an HQ3 and and HQ4 in other cities that have the specialized labor they need. Others have noted this possibility: Richard Florida tweeted to this effect last week.

There’s another good reason to pick multiple cities. If a single winner is announced, and its competitors are dismissed, then Amazon’s negotiating position becomes much weaker. A city may not be able to deliver everything that’s promised (especially over time), and local political demands for Amazon to provide compensating benefits to the community in exchange for its subsidies are likely to escalate. Having mulitple winners will allow Amazon to continue to keep each of them honest. This has been a lucrative game for Amazon: there’s no reason to end it now.

We’re a bit wary of reading too much into the list of finalist cities. Some may be included for public relations reasons, and as we predicted many of the choices reflect the desire to maximize the companies negotiating leverage when it comes to subsidies (hence the choice of three locations in metropolitan Washington). That said, it seems likely that some of the smaller cities on the finalist list (Pittsburgh, for example), may be there not because they are candidates for HQ2, but because they may be a logical HQ3 or HQ4 for a particular specialization (like robotics).

Having several smaller HQ2, HQ3, and HQ4 locations would also work to Amazon’s advantage in minimizing its impact on housing affordability. Much has been made of a likely “winner’s curse:”  in addition to giving away substantial public subsidies, a city that suddenly got 50,000 highly paid jobs would likely see a substantial escalation of housing prices.  Splitting HQ2 among several different cities would lessen housing market dislocations (and not incidentally, keep alive the prospect of future competition among these “winning” cities for subsequent job expansion).

At this stage, it’s very much in Amazon’s interest to make everyone think that there’ll be just one big “winner-take-all” first prize in the HQ2 sweepstakes. A big prize is designed to elicit very large subsidy offers and other concessions, like New Jersey’s $7 billion incentive package. But like any reality-TV show, there’s nothing that stops the producers from injecting a late-in-the-game plot twisting rule change. Don’t be surprised if later this year, Amazon announces that its going to have more than one HQ2, and that the corporate functions it moves to those places don’t have a close connection to a city’s established industrial and technological specializations.