By Nicholas Samuel
Director of Research
There are realities for every firm that allow for a natural process of filtering out site location candidates that would not be able to meet labor, site, or market access requirements. Likewise, firms typically have preferences for the kinds of communities and associated amenities that they place value on. Each company has its own idiosyncratic requirements and preferences that determine location decisions, making it difficult for outside analysts to pinpoint how a company like Amazon will place value without consulting the company’s decision makers.
We can take the stated requirements of the RFP to heart and apply some of the criteria to determine which communities would automatically be in that bunch. That is an exercise that has been taken on by a number of authors (with particularly good rationales provided here and here) and has offered up a reasonable list of metros that are likely to find themselves among the top few candidates Amazon considers. Yet, one factor that shortens the candidate list early on in the RFP is the “preference for…Metropolitan areas with more than one million people.”
The majority of metropolitan areas in the US have fewer than 1 million residents and many under that threshold are just as able to attract talent to help fulfill the labor requirements for HQ2. Labor availability is the biggest consideration to be made for any office location, especially considering that Amazon predicts that the HQ2 will hire 50,000 someday. If we estimate that a third of the total potential employees would be need a background in computer science or software development and that the hiring of these technically skilled workers would take place over approximately 15 years, we can say that Amazon would likely need to hire about 1,000 software development engineers per year. The company would also need about three times as many candidates as hires to meet their needs. Therefore, the real limiting factor is if a community has a supply of about 3,000 of these workers currently or, to be safe, 9,000 for the first few years of operation.
If we take this, and the international airport requirement (which is a high hurdle for many smaller MSAs), as the main limiting factors instead, we find that there are four metro areas that make the cut. Boulder, CO, Durham-Chapel Hill, NC, Honolulu, HI, and Harrisburg-Carlisle, PA are the four metro areas under 1 million people that meet these two requirements (Provo, UT just barely missed the cut since a drive to Salt Lake City’s international airport was just over the 45 minute timeframe set by Amazon’s RFP).
Distance takes Honolulu off the shortlist, while Harrisburg is a less likely contender given the lack of a large nearby educational/research institution. Boulder and Durham-Chapel Hill would both be sensible and intriguing choices for Amazon and can offer great access to tech talent, well-regarded educational and research institutions, and stable and growing communities. A recent report by Indeed’s Hiring Lab ranks Boulder and Durham-Chapel Hill’s research triangle neighbor Raleigh as the fifth and eleventh most similar metros, respectively, to San Jose according to their mix of tech-related jobs. In comparison, Amazon’s HQ1 Seattle ranks fourth on this list.
Many of the requirements and preferences listed in Amazon’s RFP arise from the realities of doing business for a company like Amazon. Yet, the preference for a large MSA tells us a bit more about what the firm values than many of the other items referenced in the RFP. In some sense it tells us that Amazon is betting on a market that is going to be even more urbanized, instead of remote, and a base of talent that will continue to be drawn to the largest urbanized areas instead of mid-sized, and sometimes economically more advantageous, metros. The upshot of this is that economic developers of small to mid-sized metros will need to find better ways to market their assets and economic benefits to companies, because it’s a tough path to the second round when you aren’t even allowed to audition.
