Since Amazon announced last year that it is going to build a second corporate campus, cities—238 of them in North America, in three countries—quickly started courting the company. They scrambled to propose the most generous package of financial incentives they could muster, in hopes of luring the online-retailing and cloud-computing giant.

On Thursday, Amazon announced that it had whittled its list down to 20 finalist cities spanning the country, from Los Angeles to Austin to Boston and Miami. What does the future hold for the lucky winner? In Amazon’s request for proposals, it dangled the promise of hiring up to 50,000 full-time employees (at an average salary of more than $100,000 a year) over the next 10 or 15 years, and spending $5 billion in the process of executing the project.

For cities, that means an influx of smart young workers (who then will spend a lot of their paychecks at existing local businesses) and the increased likelihood of long-term economic stability—things that many places want desperately. Indeed, Chicago (still in the running) offered to let Amazon keep $1.32 billion in taxes paid by its own employees. And the mayor of Fresno, California (no longer in the running), told Amazon the company would pay taxes alright, but 85 percent of those taxes would be poured into an “Amazon Community Fund” that would be half-controlled by Amazon’s own executives. Several other cities’ proposals haven’t been made public.

There are no doubt perks to being home to a huge, stable employer. But the way most cities pursue that goal—by offering to forfeit enormous amounts of tax revenues—produces outcomes that have worried many economists for years.

And while the competition that Amazon has put on is unique in its scale and fanfare, forgoing tax revenues is an all too popular way that cities try to attract even much smaller companies. In a 2012 series on incentives, The New York Times concluded that states, counties, and cities provide companies $80.4 billion per year through these arrangements. The practice of luring businesses with incentives such as tax breaks, grants, free land, and low-interest loans has become so ingrained it’s now a rote drill, complete with specialty consulting firms that help companies negotiate such giveaways.

Amazon’s beauty contest commenced shortly after Wisconsin signed a deal with Taiwan’s Foxconn, which makes products for Apple, the most valuable company in history. Wisconsin agreed to provide the Taiwanese firm with up to $3 billion in subsidies in exchange for building a new plant that would employ up to 13,000 people, at an estimated cost to the state of about $230,700 per employee if the deal reaches certain employment benchmarks. (That cost estimate may turn out to be low. The actual total incentives Wisconsin put forward now appear to include hundreds of millions more.)

“Alabama does it. Mississippi does it. Mexico,” an Ohio state legislator said to me one day back in December 2014, speaking of tax breaks and other incentives that aim to convince businesses to move to, or create, new facilities and jobs within their borders.

“But aren’t you just racing to the bottom?” I asked, thinking of dollars that would not find their way to schools, infrastructure, health care.

“We have to do it,” he answered. And he didn’t seem happy about it.  

Jeffry Harris, the chief of the Area Development Foundation of Knox County, Ohio, the umbrella economic-development agency representing county and town governments, is an incentives skeptic who says he has “been watching the Amazon thing with a smirk.” “I have seen communities across Ohio make very lucrative deals to get companies to move into their community and frankly believe those communities many times overpay for that development,” he told me. Local officials want to be seen as doing something to create jobs, and to project a pro-business image. Spreading taxpayers’ dollars to prospective businesses is one easy way to do it.

But Harris believes that’s unsustainable. Better, he argues, to put more effort into community development, the school system, the workforce, the roads: “What are we doing to address the overall aesthetics of our community so when I get a Lansing, Michigan, guy coming in, downtown looks sharp? If I have a plant manager who comes through with his wife, and she says, ‘This place is terrible,’ we’ve lost that opportunity.”

Harris does not eschew all incentives. He’s written six tax-break deals for companies that, for example, want to move to town and rehab an old building. But he avoids those that seem more interested in an incentive package than in becoming part of the community.

The question of whether, or how much, incentives actually spark a community’s economic growth is still unsettled. That’s partly because coming to any bottom-line answer is extremely difficult given all the possible variables in any scenario. “The overall conclusion is that effectiveness is there,” says Peter Fisher, a professor emeritus at the University of Iowa and the research director of the nonprofit Iowa Policy Project. “But it’s pretty small, and small enough that incentives end up being a very costly strategy.” In his opinion, far too many state and city boosters indiscriminately spray financial giveaway packages, which ends up costing them more than it should.

Fisher points to the example of Iowa, which has succeeded in drawing tech giants such as Google, Facebook, Apple, and Microsoft into its borders. It has given away generous incentive packages to support the creation of server farms. But such facilities can sometimes employ fewer than 100 people once they’re built, meaning the state and communities effectively pay hundreds of thousands of dollars per job gained.

“That makes no sense,” Fisher says. “It really does nothing for the high-tech sector in Iowa—it’s just a name on a big building.” The main reasons tech giants have built those farms is not the incentives. Rather, it’s Iowa’s abundant, cheap energy (thanks partly to the growth of wind power), lots of land, and few natural disasters.

Fisher notes that whether it’s a server farm, a manufacturing plant, or a new headquarters, states and communities often neglect to consider, or fail to publicize, ancillary costs beyond the actual incentives. In 1983, for example, after the little town of Wellston, Ohio, and the state, used tax breaks to attract a Jeno’s frozen-pizza plant from Minnesota, the town’s sewer system nearly collapsed while handling 400,000 gallons of pizza-ingredient sludge emerging from the factory. So the state had to use a federal block grant of over $500,000 to bail out the company and the town.

Most secondary costs are more prosaic. New business brings new growth and demands, which may be a net good, but also include costs like new roads, more children in the school system, more housing, more waterworks. All that has to be paid for, and when tax revenues are cut via incentives, there’s less money to pay for it.

And incentive deals can go sour in more dramatic ways. Companies have a long history of reneging on promised jobs and development, and states can fail to include strict job targets in their packages. Louisiana, for example, has lost over 36,000 manufacturing jobs even as it continued doling out expensive incentive packages. Chiquita Brands was seduced away from Cincinnati to Charlotte with a $22 million package including rebates on state payroll taxes, and millions in grant money. Three years later, it moved out of Charlotte, leaving city leaders to question the whole idea of incentives.  

Even those who aren’t so skeptical of incentives are aware of their limitations. The truth is, local and state tax structures aren’t really so critical to most businesses, especially large ones, for which such taxes can be a relatively small part of their total costs. “Economic development is a useful tool, but I would never advise a company to use economic incentives to ID a longlist of candidates,” says Josh Bays, a principal at Site Selection Group, a location-advisory firm. He likes incentives, and takes a dim view of academic research suggesting they don’t do much to improve the lives of cities. Even so, he advises clients to consider more-important attributes of a location, such as transportation costs, raw-materials availability, and the size and quality of the area’s workforce, before thinking about incentives. Only then should they be used to break a tie between one or more finalists.

The present state of affairs is exactly what policymakers in Washington, D.C., were worried about 21 years ago. “We’re here to talk about a war, an alleged war among states and localities right here in the U.S.A.,” Alice Rivlin, then the Vice Chairman of the Federal Reserve Board, said in her keynote address at a conference about the issue in the spring of 1996.

The conference was sparked by a 1994 essay, “Congress Should End the Economic War Among the States,” by two Minneapolis Fed staffers, Arthur J. Rolnick and Melvin L. Burstein. “While states spend billions of dollars competing with one another to retain and attract businesses,” Rolnick and Burstein wrote, “they struggle to provide such public goods as schools and libraries, police and fire protection, and the roads, bridges and parks that are critical to the success of any community.” America is one economic family, they argued, and unseemly competition to attract new businesses, or to retain existing businesses located within any particular state or city, “undermines the national economic union.”

One advantage of the American system, with powers divided between states and a national government, Rolnick and Burstein pointed out, is that states are free to compete with each other by trying different taxing and spending allocations. For example, one state may tax a bit more and provide more public goods—better schools, cheaper health care, smoother roads, more-pleasant parks—in return. Another state may tax less, and spend less, on such public goods. People and businesses could then choose where they preferred to live and locate. Those choices, in theory, help select which balance of taxing and spending emerge as the best.

But this competition is perverted when it’s applied to individual corporations. If one state fends off a poaching attempt by another state by offering bigger payouts, Rolnick and Burstein wrote, “competition has simply led states to give away a portion of their tax revenue to local businesses; consequently, they have fewer resources to spend on public goods, and the country as a whole has too few public goods.” If a state successfully draws a business into its borders through tax benefits, they go on, “there will be fewer public goods produced in the overall economy because, in the aggregate, states will have less revenue.”

Rolnick and Burstein believed there was a solution to all this: federal legislation. Congress, using its power to regulate interstate commerce, could change the way relocating companies get taxed (taxing them based on the estimated value of the benefits, perhaps) or threaten to withhold federal funds from states that try to poach businesses by offering to give up taxes. What with 238 cities having vied for Amazon’s favor—and Amazon now having made a show of narrowing its list down to 20 cities—such legislation seems mirage-like at the moment.