Blog | IMPLAN

An Apples-to-apples Look at Amazon HQ2 Candidate Cities

Written by Tim French | February 14, 2018

Tale of twenty cities

Only weeks ago Amazon announced the top 20 candidate cities still in the running to capture the lofty honor of being home to Amazon’s second headquarters—or ‘HQ2’. Amazon culled their short list from a whopping 238 bids since the deadline for submissions in mid-October 2017. Factors affecting the decision process included proximity to a city center, quality of commute for potential workers, proximity to an airport, demographic diversity, and an emphasis on revitalizing an existing building rather than moving into something newly-constructed.

RFP responses landed in Amazon’s inbox in a wide variety of shapes and forms from Boston’s 218 page report to the City of Frisco, Texas’s enthusiastic video—other candidates were decidedly more guarded with their submissions. I don’t envy Amazon its task to make a final decision, but I do know of at least one way to level the playing field and see what sort of economic impact opening a headquarters in any one city might have.

Amazon’s original RFP gives us enough details about employment, employee compensation, and a few other economic factors to build a list of monetary shocks (or direct impacts) that we can model in each city’s metropolitan statistical area (MSA). Results with this approach may change with more detailed information (like how the 50,000 employees break out into job types or specific compensation levels, for example).

I’m taking a very broad-strokes approach to comparing the candidate cities. However, anyone with more detailed information about the direct impacts and the right economic modeling tool could explore more granular, detailed results. For curiosity’s sake, here’s what I modeled in each city:

What it takes to run a headquarters

Here’s what we know Amazon plans to spend during construction starting in 2019 and subsequently during HQ2 operations:

Construction

Let's assume two years for each construction phase.

  • Phase I (starting 2019): $450,000,0000 (RFP lists $300,000,000-$600,000,000)
  • Phase II Building: $837,000,000 ($600,000,000-$1,260,000,000)
  • Phase III Building: $1,622,500,000 ($1,260,000,000-$1,985,000,000)

Operations

  • 50,000 Employment
  • $100,000 average compensation per employee
  • $1.4 billion operational expenditures (utilities and maintenance)

And here’s what I found:

Cities ranked by total impact

Total impact includes the direct impact of the known economic events that I’m modeling as well as the indirect and induced effects. Broadly speaking, you can think of indirect effects as the supply chain. The impact of local industries buying goods and services from other local industries. The cycle of spending works its way backward through the supply chain until all money leaks from the local economy (due to taxes, savings, imports from outside the region, etc.). Induced effects are the response by an economy to an initial change (Direct Effect) that occurs through spending of labor income (employee compensation + proprietor income) throughout that region after taxes, savings, commuting, and imports have been accounted for.

City Total Value Added Total Output
Toronto $80,268,623,744 $82,159,539,571
Atlanta $85,597,226,718 $75,025,781,810
Dallas $84,077,877,067 $73,495,393,649
Nashville $82,686,085,348 $71,882,183,928
Austin $82,256,135,490 $71,574,167,796
Chicago $83,868,806,546 $71,237,728,142
Miami $81,760,324,185 $71,173,394,717
Indianapolis $82,422,094,707 $70,906,416,353
Denver $81,654,926,288 $70,161,891,650
Pittsburgh $83,090,441,134 $70,152,561,498
Philadelphia $83,134,769,316 $69,096,469,856
Columbus $81,638,769,688 $68,367,486,498
Montgomery County, MD $81,689,906,169 $67,586,614,804
Boston $81,343,342,888 $66,479,836,634
Raleigh $79,507,310,266 $65,736,928,294
Los Angeles $80,007,273,668 $65,175,110,040
Newark $80,464,271,946 $64,989,191,930
Northern Virginia $75,981,861,566 $57,435,385,632
New York City $75,894,276,260 $55,082,849,959
Washington D.C. $62,963,024,653 $36,414,373,730

 

Cities ranked by employment

Employment in IMPLAN includes full-time, part-time, and seasonal workers and therefore does not represent full-time-equivalents (FTEs). IMPLAN employment includes wage and salary employees as well as proprietors (self-employed individuals and unincorporated business owners).

City Indirect Employment Induced Employment Total Employment Total Employee Compensation
Austin 59,197 303,974 414,670 $514,568,278,854
Toronto 56,153 276,293 383,946 $559,884,131,293
Miami 46,568 326,272 424,340 $563,680,398,622
Dallas 43,104 312,377 406,981 $564,408,170,716
Indianapolis 42,463 327,600 421,564 $563,698,546,506
Nashville 41,781 308,384 401,665 $563,484,679,250
Atlanta 41,476 346,404 439,381 $515,443,359,897
Raleigh 40,194 295,222 386,916 $562,252,700,622
Denver 39,011 306,921 397,433 $564,093,775,466
Los Angeles 37,376 257,115 345,991 $562,324,548,238
Montgomery County, MD 36,500 285,559 373,559 $563,773,543,692
Columbus 35,167 314,559 401,225 $562,705,045,001
Chicago 34,243 307,962 393,704 $565,048,025,992
Pittsburgh 32,066 330,088 413,654 $564,637,914,457
Newark 30,569 260,449 342,518 $563,249,479,124
Boston 30,027 270,391 351,919 $565,219,236,915
Philadelphia 29,560 291,390 372,450 $564,420,270,619
Northern Virginia 27,224 219,715 298,439 $560,571,359,333
New York City 25,345 188,268 265,113 $561,283,942,889
Washington D.C. 22,560 84,115 158,175 $555,837,625,581

 

Takeaways

You might be surprised to see that New York City and Washington D.C. are not the biggest apples. That's partly because the cost of living in those metropolitan areas is generally higher than other parts of the country. This means that the jobs supported in the indirect and induced impacts are lower compared to cities like Austin where wages can be lower. 

And there are other factors to consider—specifically, feasability. New York City and Washington D.C. already have robust logistical infrastructures in place which might benefit a corporate headquarters and more than counterbalance the weight on the potential economic impact.

Ultimately, Amazon's HQ2 is going to have a significant economic impact no matter where it lands.

A note about methodology

Arguably the best way to measure the sort of impact comparison that we’re looking at here would be with a computable general equilibrium (CGE) model. However, since Amazon reported the impacts of its HQ1 based on the results of an input-output model, I felt it best to follow suit—if for no other reason than to maintain thematic consistency (input-output modeling is not a reliable forecasting tool so the results we’re looking at are purely speculative).

IMPLAN is actually really well suited for this type of analysis! However, since HQ2 will be so large, it may have some pretty big effects of other types that IMPLAN can't get at (many of which CGE also wouldn't get at). The things it will miss (that should be considered separately if possible) are things like:

  • Environmental or Social issues (traffic, gentrification, wage effects, etc.)
  • Investment the city would have to make in infrastructure to support it (new roads, pipes, traffic lights, etc.).
  • What amount of money each city is putting up as incentive and what the city would have done with those funds otherwise (e.g., are we skimping on improving our roads and schools so we can incentivize Amazon?)

The IMPLAN modeling system can examine U.S., state, county, multi-county, city level, and international economies to assess the total economic impact of a variety of situations; a capital investment in a project, a business change, the economic consequences of natural disasters, the current value of a business to a regional economy, or the jobs supported by a local supplier.

By constructing social accounts that describe the structure and function of a specific economy, IMPLAN allows users to create localized models to facilitate investigation into the potential consequences of projected economic activities. IMPLAN allows you to identify and quantify the direct, indirect, and induced impacts of a business/supplier.

IMPLAN databases are constructed exclusively by IMPLAN’s team of PhD economists. All IMPLAN datasets use a set of annual national I-O matrices as the background framework for the software and database interaction. All data files include numerous economic and demographic variables at a 536 Industrial sector level. Data variables include employment, value-added (GDP), government and household purchases, taxes, and more.