Flawed “cost-benefit” analysis is being used to justify a massive Amazon subsidy, watchdog leader says

August 4, 2022

Testimony of Greg LeRoy Executive Director, Good Jobs First

To the Niagara County IDA Board Regarding the

“Cost-Benefit Calculator” Used to Justify IDA Subsidies for Amazon Project


Good Jobs First today released critical testimony on a cost-benefit “calculator” being used to justify a $124 million subsidy for an Amazon warehouse in Niagara County, New York: 

“I direct Good Jobs First, a Washington-based, non-profit, non-partisan watchdog group on economic development incentives since 1998.

I question the “cost-benefit calculator” that the NCIDA is using to justify its proposed $124 million subsidy for the Amazon warehouse.

The study fails to state its underlying assumptions, which are critical to the credibility of any cost-benefit analysis. Nor does it name any software products it may have used to derive its estimates.

There is no indication that the cost-benefit calculator accounts for job destruction in the bricks-and-mortar retail sector caused by the rise of e-commerce/Amazon. That job destruction will offset warehouse job gains to an unknown degree. Just because people have another way to shop does not mean they have more money with which to shop.

Related: the calculator projects an increase in local property tax values of $48.3 million. Again, underlying assumptions are not stated. As the consulting firm Civic Economics has documented: retail property values, occupancy rates and tax assessments all decline as e-commerce sales increase (with Amazon comprising almost half of e-commerce).

Also at risk are residential property values along the routes serving the warehouse. Having hundreds of trucks traversing those streets every day may harm home resale values, as air quality declines and noise increases. Lower resale values mean lower assessments and lower property tax revenues.

The calculator uses the words “local” and “state” to distinguish benefits, but it makes no distinctions between Niagara County and Erie County. Given the fact that Erie County’s population is about four and a half times that of Niagara County, it is reasonable to assume that many if not most of the local job takers will be Erie County residents.

Indeed, Erie County may well be the big local winner if Niagara County subsidizes Amazon: Erie could get many if not most of the jobs (and also the property-value benefits) while suffering none of the PILOT tax revenue losses, and little of the concentrated air pollution, noise, or road wear.

The study apparently assumes a constant level of 1,000 permanent employees over 15 years. This may not be a valid assumption, given Amazon’s aggressive moves towards automation.

The study projects 527 “ripple effect” permanent jobs. A share of “ripple effect” jobs are normally public-sector jobs. But for the first seven years, the PILOT abatement rate is 90%, and it does not even reach 50% until year 11, so the project should generate far fewer public-sector jobs than it would without the abatement. Does the calculator reflect that? We cannot tell.

The most troublesome aspect in the calculator is the mixing of apples with watermelons. Central to its finding that local benefits will outweigh local costs is the juxtaposition of payroll earnings with incentives.

This is not a valid comparison. The PILOT incentive equals public dollars out, whereas earnings (i.e., payroll) do not equal public dollars in. Someone earning $32,640 here pays state income tax of 5.2%. So only 5.2% of the projected earnings can be correctly counted as a revenue benefit against the PILOT incentive cost.

And it is a huge distortion: projected payroll of $1.34 billion equals 91.3% of the calculator’s total projected benefits.

The calculator also apparently double-counts $69.6 million in project benefits. That is how much in state sales and income tax revenue it attributes to the temporary (construction) and ongoing (permanent) jobs. But there is no notation that it subtracted that sum from the payroll estimates. But of course, all of the new income and sales tax revenue can only come from those payrolls, so the calculator apparently counts the tax-spending share of workers’ paychecks twice.

So for all these reasons, in our opinion, this is not a fiscal break-even analysis, or a taxpayers’ cost-benefit analysis. It is a misleading public relations document.”