Critique of Sage Policy Group "Economic Contributions to the Potential Amazon HQ2 in Maryland"
By: Jacqueline Coolidge[1]
The only publicly available analysis of the purported benefits to Maryland, if Amazon were to locate its second headquarter ("HQ2") in the state, was one commissioned by the State of Maryland and Montgomery County Maryland and undertaken by Sage Policy Group, Inc. in February 2018.
The analysis is based on the IMPLAN economic modeling software, which is a reputable and commonly used source for estimating the "multipliers" (additional economic activity and employment) that might be generated by a particular size and type of investment in a particular location. Sage's report concluded that the state of Maryland (including Montgomery County but also several other neighboring counties) would benefit with about 50,000 jobs associated with the construction of the HQ2, which would, once fully operational, "annually support more than 101,000 jobs in Maryland ... and more than $17 billion in economic activity." (Sage, pg. 4).
Subsequently, the state of Maryland and Montgomery County decided to offer a package of incentives to Amazon of about $8.5 billion to secure a decision to site the HQ2 in Montgomery County. The figure includes about $2 billion in infrastructure paid for by the state and about $6.5 billion in “credits, incentives and grants,”[2] at least some of which would be contingent on employment. The size of the package is apparently the biggest on offer to Amazon in the country, the next-largest being that of New Jersey at about $7 billion, and many times larger than anything ever offered in Maryland previously.[3]
According to the Baltimore Sun:
The economic development deal, which Hogan called the PRIME Act — named for Amazon’s membership program — would provide Amazon incentives if it creates at least 40,000 jobs that pay an average of $100,000 a year. The proposal would provide a $10 million-a-year grant out of the state’s Sunny Day fund, it would give a state sales tax exemption for construction materials used to build the project, and it would create a new, annual tax credit equal to 5.75 percent of each new job’s wages. The company could apply for that tax credit every year.
Legislative analysts say the package would cost state and local governments $6.5 billion over the next 35 years in lost tax revenue and increased costs.[4]
Does the offer of the incentive package make sense? Maybe; maybe not. First we need to consider whether the assumptions and analysis presented by Sage are sound and reasonable. Even more important is to consider the cost side of the investment, which the Sage report did not do at all. The Sage report says it did not take into account the incentive package; unfortunately the details of the “legislative analysts” assessment of the costs are not available for public scrutiny.
Even before examining the Sage report in detail, I would ask whether the IMPLAN model would be likely to yield an accurate estimate of the associated economic activity and employment to be generated. The model was developed in the 1970s and has been updated over time, but we should keep in mind that a "second headquarters" of a corporation of any size does not have many precedents. The model therefore has little data that is relevant for forecasting. Will HQ2 be mostly a duplicate of the Amazon HQ in Seattle? And will it operate in parallel "into the indefinite future"? Or will it be used as a very large bargaining chip to play off one jurisdiction (i.e., Seattle, WA) against another one (MoCo MD, on the assumption that it wins the bid)? We should all be familiar with threats by large corporations to move away if they don't get yet more tax and regulatory breaks in the future. Such threats become much easier to implement (and would therefore be more credible) if a corporation indeed has two operational headquarters open that it can use to shift resources relatively quickly.
Sage describes some of the assumptions used in the report in the Introduction. The report quotes Amazon's press release from January, which stated that Amazon "plans to invest more than $5 billion and grow this second headquarters to accommodate as many as 50,000 high paying jobs." (Sage pg. 4) Let's be clear: this is a press release, not a legal contract, as "as many as" is usually understood to mean "at most." So our first question is whether we should use 50,000 as an accurate prediction of the number of jobs at the HQ2. History across the US and across the world shows that pre-decision announcements by large investors who are deciding on an investment location (and playing off different jurisdictions in the hope of maximizing the "incentives" offered to them) rarely come through as promised.[5]
Sage then says "This analysis endeavors to quantify additional jobs, compensation, and local business sales associated with both a partially and a fully built-out HQ2." (Sage pg 4). In other words, it is presenting the "benefit" side of the equation, but there is no mention of the cost side (see below). Sage also assures us that "fiscal impacts are generated within our IMPLAN model and verified for accuracy by hand using effective tax rates calculated with data made available within Maryland Comptroller's Certified Annual Financial Report." (Sage pg 4). Unfortunately, the term "effective tax rates" has multiple definitions, and Sage never tells us which one it is using.
The report notes that "Amazon expressed in its initial RFP plans to invest more than $5 billion in capital investment over the first 15 - 17 years of HQ2. Approximately $1.26 - $1.99 billion of that investment is anticipated to fall under the umbrella of the initial three development phases." (Sage pg 5). The report offers no information about how long phases I - III are expected to last. There is mention of "Phase IV and any additional phases of capital investment needed to bring HQ2 to full build out..." (Sage pg 5). Sage gives us plenty of "outputs" in terms of "Jobs (Full Time Equivalents)”, associated "Employee Compensation" and "Business sales" as estimated by the model, but no details about the input assumptions.
The report also states "jobs supported outside the study area, including in the District of Colombia and in Northern Virginia have not been estimated and are not included in our findings." (Sage pg. 7). I would want to know how they handled the analysis within the model for a large investment, which would be expected to have substantial spillover benefits in DC and Northern Virginia. The model normally provides estimates for an entire economic region, so how did Sage exclude DC and NoVa? They never offer a word of explanation.
In estimating the ongoing impact of HQ2 during its operational phase, the report only bothers to disclose three assumptions:
HQ2 reaches the stated goal of 50,000 employees earning an average of $100,000
HQ2 matches the Seattle Headquarters total number of annual hotel nights by visitors ...
The amount paid into the region's public transportation system through employees' transportation benefits matches that of the Seattle Headquarters..." (Sage pg 8)
The report say "In total, HQ2 will support more than 101,000 annual jobs within the study area. This tally includes both direct and secondary jobs." (Sage pg 9). They further estimate that the average such job will pay almost $76,000. Is that realistic? In the appendix, Sage notes that "IMPLAN also provides the user with a choice of trade-flow assumptions, including the modification of regional purchase coefficients, which determine the mix of goods and services purchased locally with each dollar in each sector. Moreover, the system also allows the user to create custom impact analyses by entering changes in final demand.” It then asserts Sage is "uniquely qualified to develop data and factors tailored to this project, and, where appropriate, overwrite the default data contained in the IMPLAN database." (Sage pg 12). Unfortunately, they offer no further information about what choices they made; what modifications they made; and what "changes in final demand" they made or why.
For "Fiscal Impacts" the report offers estimates of $280 million annual at the "county level" and almost $483 million at the state level (Sage pg 9). They talk about "effective tax rates" without bothering to specify whether they are "average" or "marginal" effective tax rates nor what definitions or methodology were used. The differences are often substantial. [6] Further, the report only offers an estimate of "tax receipts" but fails to offer any information whatsoever about the other side of the ledger for "fiscal impacts": the expenditure side. The report offers no information whatsoever about the need to expand infrastructure to accommodate a campus for 50,000 workers (will the Maryland $2 billion investment offered as part of the package suffice?); nor the housing for the proportion of the workers at HQ2 who might not already live in the area (nor what that proportion might be); nor the associated need to build and staff schools and other social services; nor the cost of the congestion and wear-and-tear on existing infrastructure.
Worst of all, the report is perhaps misleading for many readers in that it forecasts "impacts" in terms of business "sales" estimates (the much-touted $17 billion figure), but not "value added." According to an analysis published by Brookings Institute, “[Amazon] employs tens of thousands of workers in Seattle and claims an economic impact of $38 billion between 2010 and 2016."[7] Estimates of this nature vary considerably, and it’s probably safe to assume if Amazon itself is claiming $38 billion over a six year period, that it’s probably the most flattering estimate that they could tout with any credibility. So how does Sage get away with an analysis claiming that MD will enjoy $17 billion per year?
In short, the report is maddeningly vague about its sources of data, assumptions, and what changes were made to the standard IMPLAN model. It presents only the "benefit" side of the ledger and ignores the cost side. Legislative analysts reportedly estimated total costs at $6.5 billion, but we don’t have enough detail to judge the soundness of that figure either. The discussions about the incentive package were secret, as are the fine details of the incentive packaged offered. There has been no discussion about the distributional effects of HQ2 (e.g., what is likely to happen to housing prices and property taxes for households on fixed incomes). The secrecy (a typical demand of the investor, who benefits in the negotiation process with multiple potential host jurisdictions competing against each other) allows for bad policy.
In my previous work for the World Bank Group’s Foreign Investment Advisory Service[8], our “best practice” advice to governments who were interested in attracting investment was to avoid offering special “incentives” to large investors (because politicians typically want to “win” the investment and investors play off jurisdictions like auctioneers encouraging bidding wars that lure participants into bidding more than a particular item is worth). That was always an uphill (and usually losing) battle for us, so our fallback “good practice” advice has been to insist on transparency in the design of the incentive package, robust debate about assumptions used in the analysis and careful, explicit consideration of the likely “winners” and “losers” associated with the investment and the incentive package. The incentives should be accounted for in detail as “tax expenditures” to allow for both future “lessons learned” basic political accountability. Brookings recommends the same to U.S. jurisdictions: “localities must commit to making incentives information publicly transparent, and then rigorously evaluate their impact on firm outcomes to determine what works.”[9]
The offer made by the State of Maryland and Montgomery County is now a fait accompli. Although I don’t have enough information or data to make my own estimate of the net benefits of an Amazon HQ2, my hunch is that our politicians got suckered into bidding more than the deal would be worth to us. My hope at this point is that we will lose the bid, but that the new County Council and State legislators will agree that in the future, such bids must be considered on the basis of full transparency and robust debate. Investors who insist on secrecy should be politely shown the door.
[1] Retired economist from World Bank, former Lead Investment Policy Officer
[2] http://www.businessinsider.com/amazon-hq2-tax-incentives-maryland-2018-4
[3] http://www.businessinsider.com/amazon-hq2-tax-incentives-maryland-2018-4
[4] http://www.baltimoresun.com/news/maryland/politics/bs-md-amazon-package-passed-20180404-story.html
[5] See, for example, Devereux, M.P., Griffith, R & Simpson, H. (2007). Firm location decisions, regional grants and agglomeration externalities. Journal of Public Economics, 91, 413-435; Klemm, A., & Van Parys, S. (2012). Empirical evidence on the effects of tax incentives. International Tax and Public Finance, 9(5), 393-423.
[6] https://www.taxpolicycenter.org/briefing-book/what-difference-between-marginal-and-average-tax-rates
[7] https://www.brookings.edu/blog/the-avenue/2017/09/08/which-cities-are-well-positioned-to-land-amazons-hq2/
[8] Later called Investment Climate Advisory Services; see for example James, S. (2009). Tax and non-tax incentives and investments: Evidence and policy implications. Washington, DC: Investment Climate Advisory Services of the World Bank; Holland, D., & Vann, R. (1998). Chapter 23. Tax law design and drafting. In V. Thuroyi (Ed.). Income tax incentives for investment (pp. 789-867). IMF.
[9] https://www.brookings.edu/research/examining-the-local-value-of-economic-development-incentives/