Amazon.com Inc. CEO Jeff Bezos walks onstage for the launch of the new Amazon Fire Phone in Seattle.

Amazon.com Inc. Chief Executive Jeff Bezos late last month cracked a bottle of champagne at the top of a 300-foot wind turbine for the opening of the Amazon Wind Farm in Scurry, Texas. This project, developed by Lincoln Clean Energy, will now provide Amazon with clean energy to power the equivalent of 90,000 houses. But the project itself is part of a controversial economic development project that offers policymakers insights into whether state and local tax incentives are worth offering to companies shopping around for new places to invest and do business. The lessons could be particularly relevant to states and localities that late last month also finalized their economic development pitches to Amazon for its proposed $5 billion new second headquarters, which the technology giant says will result in the creation of 50,000 new jobs.

The wind energy project that Bezos celebrated atop that wind turbine came to be through a state program called Chapter 313, which I examined in a recent working paper. A more detailed look at this particular project is worthwhile because of the way it abates companies from local taxes. Under Chapter 313, individual school districts in Texas authorize a limit on local property taxes, but then these school districts are made whole by the state. So, school districts give out incentives, but the cost is ultimately paid for by state taxpayers. Estimates are that the accumulating costs of these tax abatements for this wind energy farm will total more than $1 billion per year by 2022.

The wind farm in question—originally called Dermott Wind—was authorized by the program, but not without controversy. Wind farms receive incentives at the federal, state, and local level. For this project, the major federal incentive—the Production Tax Credit—was set to expire in 2016. Like many wind projects, the developers of this wind farm sped up early construction to qualify for the production credit for that location. In short, the company locked in a lucrative federal incentive by starting construction in December 2015.

After beginning construction of this $300 million project, the company applied for the Texas Chapter 313 program. This program would authorize the $300 million project to be taxed at a value of only $30, providing more than $22 million in incentives from the Texas taxpayer, alongside an additional $1.2 million in incentives from local tax abatements. Yet the 313 program requires a company to claim that Chapter 313 is a “determining factor” in its decision to invest in the locality within Texas, and that without the program, the company would invest in a location outside the state. So, Lincoln Clean Energy needed to apply for a federal incentive claiming it was already building in Texas and, at the same time, claim to Texas that it could move elsewhere if it didn’t qualify for 313 treatment.

These mutually exclusive claims did not go unnoticed by the Texas Comptroller’s office, where formal certification indicates even further evidence of building on the site prior to receiving incentives. Yet despite this evidence that the project was locked in this location, this incentive was authorized by the school district and certified by the Texas Comptroller. The paper trail, including the federal Production Tax Credit, indicates the company was coming to Texas whether or not this 313 incentive was authorized.

What does this odd case tell us about economic development? Let’s look at that other and much-larger project in which Amazon’s Bezos has a very direct hand. Amazon late last month also closed its public call for proposals from cities and states around the country for a giant new second headquarters. This direct appeal by the company for public subsidies led to shocking offers of $7 billion in tax and other public incentives from New Jersey and a number of playful stunts, ranging from the mayor of Kansas City, MO positively reviewing 1,000 Amazon products to a town in Georgia offering to rename itself in honor of the company.

The Texas wind energy project—and now Amazon’s proposed new headquarters—highlights two important points. First, many of the companies applying for incentives around the country have already made their decisions about where to invest. My work on the Texas 313 program indicates that 85 percent of the companies seeking to avail themselves of the tax abatements were coming to Texas even without the incentives. Other studies find that roughly two-thirds to three-fourths of incentives are redundant, in that they are being provided even though they do not really affect companies’ decisions about where to invest and build. One study even finds that 70 percent of corporate executives who received incentives in North Carolina didn’t even know they received them. Doesn’t sound like they were very enticing.

Second, many of these state and local incentive programs are designed to provide very weak tests for providing incentives. The Texas Chapter 313 program says companies simply need to state that incentives are “a determining factor” in their decision. To qualify, companies only need to claim they have other options or that the incentive is necessary to make the project financially viable. No disclosures are required.

The bidding war for Amazon’s second headquarters demonstrates that some public officials are losing sight of many of these lessons. Many cities and states are putting on the table their best possible offers, but is that really good public policy? What if Amazon has already chosen a location—or, more likely, narrowed the choice down to a select few places and is simply taking bids to maximize its benefits? Whatever location wins the new project needs to be sure it conducts a thorough cost analysis to learn whether any tax abatements are really worth the cost.