Latest / The Fiscal Policy Podcast with Fexingo: Government Budgets, Stimulus, and Public Spending / How Government Subsidies for Amazon Backfired
Transcript
- Lucas: So there's this number that I think about a lot when we talk about government spending on economic development. In 2018, when Amazon was shopping around for a second headquarters, cities and states collectively offered something like five and a half billion dollars in total incentives. And the winner, Virginia, promised up to seven hundred and fifty million dollars in direct subsidies and tax breaks. Luna: Right, the HQ2 circus. I remember the frenzy — every major city putting together these glossy bids. Lucas: Exactly. And the rationale was straightforward: Amazon would bring twenty-five thousand high-paying jobs, plus construction, plus spin-off effects. But here's the thing — a lot of that promised job creation probably would have happened anyway, just maybe not in Arlington. A 2024 study from the Brookings Institution looked at thirty years of state-level business incentives across the U.S. and found that roughly sixty percent of subsidized projects would have occurred in the same state even without the tax break. Luna: So we're essentially paying companies for doing what they were already going to do. Lucas: That's the fear. And the worst-case examples are pretty dramatic. Take Wisconsin and Foxconn. In 2017, the state offered up to three billion dollars in tax credits for a massive LCD plant that was supposed to employ thirteen thousand people. By 2021, the project had created fewer than fifteen hundred jobs, and the plant was essentially repurposed. The state ended up paying about two hundred and forty million dollars in credits before the deal was renegotiated. Luna: Three billion dollars in promised incentives, and you get barely one-tenth the jobs. That's a staggering mismatch. Lucas: It is. And it's not just Foxconn. A 2020 analysis by the New York Times found that across the country, states and local governments spend an estimated forty-five billion dollars a year on economic development incentives. That's more than the federal government spends on the entire Department of Education. Luna: Wow. So when we talk about fiscal policy at the state and local level, this is a huge line item. Lucas: It's enormous. And the question is: does it work? The Brookings study I mentioned ran the numbers on a sample of over two hundred large incentive deals from 1990 to 2020. They compared job growth and wage growth in counties that landed big subsidies with similar counties that didn't. Their conclusion: on average, the subsidies had no statistically significant effect on local employment or wages. Luna: No effect? That's pretty damning. But I imagine the political pressure to offer incentives is intense — no governor wants to be the one who let the next Amazon get away. Lucas: That's the prisoner's dilemma of state competition. If every state is offering incentives, then no state gains a competitive advantage, but each state still loses revenue. And the companies know it. In Amazon's case, the final shortlist of twenty cities all offered packages worth hundreds of millions. New Jersey's bid was reportedly around seven billion dollars for Newark — they didn't even make the final cut. Luna: Seven billion dollars and still nothing to show for it. That's real money that could have gone to schools, infrastructure, public health. Lucas: And that's the core fiscal policy question. You know, Luna, if today's conversation helped clarify something about how these trade-offs work, it's a good reminder that listener support — even a few dollars — helps us keep digging into these numbers. If today was worth a coffee to you, you can find us at buy me a coffee dot com slash fexingo. It really does help us stay independent and focused on the data. Luna: Yeah, absolutely. We're not backed by any big institution, so that kind of direct support makes a difference. Lucas: So back to the data. Let's talk about what happens when a state does try to claw back incentives. There's a mechanism called 'clawback provisions' — basically, if the company doesn't hit its job or investment targets, the state can recapture some of the tax credits. But these provisions are often weak or unenforced. Luna: Give me an example of a clawback that actually worked. Lucas: Sure. In 2014, Oregon offered solar manufacturer SoloPower thirty million dollars in tax credits to build a plant in Portland, promising four hundred and fifty jobs. When the company only created about fifty jobs and then filed for bankruptcy, the state managed to recover about ten million dollars. That's a partial success, but it took years of legal battles. Luna: So the state got back a third of its money, but only after spending who knows how much on legal fees. Lucas: Exactly. And that's the exception, not the rule. A 2021 report from the Government Accountability Office looked at federal and state incentive programs and found that fewer than half of them had any formal clawback mechanism at all. And of those that did, only about one in five actually recovered money when targets were missed. Luna: So the incentives are effectively gifts in most cases. That's a hard pill to swallow for taxpayers. Lucas: It is. And what's interesting is that some states are starting to experiment with different approaches. For example, Michigan's 'Good Jobs for Michigan' program, launched in 2015, ties the tax credit to actual payroll — the company only gets the credit for each new job that pays above a certain wage threshold. It's more targeted and performance-based. Luna: That sounds more like a reimbursement than a giveaway. Does it work? Lucas: Early evidence is promising. A 2023 study by the W.E. Upjohn Institute found that Michigan's program generated about 1.2 dollars in new state tax revenue for every dollar of credit issued. That's a positive return. But the program is small — it covers only about five thousand jobs a year, compared to the hundreds of thousands of jobs in the state. Luna: So it's a good model, but it's not going to replace the forty-five billion dollars in incentives being spent annually. Lucas: Right. And that brings us back to the Amazon HQ2 case. Virginia's deal was actually more structured than most. It included a 'pay as you go' model where Amazon only receives credits after it has verified job creation and capital investment. Through 2025, Amazon has invested about two billion dollars in the Arlington campus and created about eight thousand jobs — well short of the twenty-five thousand target, but the state hasn't paid out the full seven hundred fifty million yet. Luna: So the state isn't out the full amount, but it's still a huge subsidy for a company that could have chosen to expand in Virginia anyway, because of the talent pool and the proximity to D.C. Lucas: That's the crux. Economists call it 'deadweight loss' — the subsidy doesn't change the company's behavior, it just transfers money from taxpayers to shareholders. A 2019 study from the University of North Carolina estimated that up to 80 percent of state incentives are redundant in that sense. Luna: Eighty percent. So out of forty-five billion, that's thirty-six billion dollars a year with basically no economic impact. Lucas: That's the fiscal cost. And then there's the distributional impact: these incentives often go to large corporations, while small businesses — which create most new jobs — rarely get similar breaks. A 2022 report from the Economic Innovation Group found that firms with over five hundred employees received 72 percent of state incentive dollars, despite accounting for only about 20 percent of net new job creation. Luna: So we're subsidizing the companies that are least likely to create the jobs we need. That seems backwards. Lucas: It is. And it's not just about efficiency — it's about fairness. When a state gives a billion-dollar tax break to a corporation, it has to cut spending or raise taxes elsewhere. Often that means cutting education, infrastructure, or health care. So low-income families end up subsidizing high-paid corporate executives. Luna: I want to push back a little. Aren't there cases where incentives genuinely tip the scales? For example, Tesla's gigafactory in Nevada. The state gave them about 1.3 billion dollars in tax abatements in 2014, and now the facility employs over ten thousand people. Lucas: That's a good example. And the evidence is mixed. The gigafactory deal did include strong clawback provisions, and employment has exceeded initial projections. But a 2020 study by the University of Nevada, Las Vegas found that the net fiscal impact was still negative — the state lost more in tax revenue than it gained from new economic activity, because the abatements were so generous. The study estimated a net loss of about forty million dollars over the first five years. Luna: So even a 'success story' can be a net loss for the state budget. Lucas: Exactly. And that's the pattern: even when job creation is real, the tax breaks are often so large that the state never recoups its investment. The exception might be when the incentive is small relative to the economic impact — for example, a training grant that costs a few hundred thousand dollars but brings in a factory with a million-dollar payroll. But those aren't the headline deals. Luna: So what's the alternative? If states want to attract business, what should they do instead of these megadeals? Lucas: A growing consensus among economists is to focus on 'public goods' rather than firm-specific subsidies. Invest in high-quality schools, reliable infrastructure, a skilled workforce, and a predictable regulatory environment. Those benefits are available to all businesses and residents, and they don't distort competition. A 2024 paper from the International Monetary Fund found that for every dollar spent on public education, the long-run return to state GDP is about three dollars — far higher than the return on targeted corporate subsidies. Luna: So we're back to basics. Invest in the things that make a place attractive generally, rather than trying to pick winners. Lucas: Exactly. And that's a more sustainable fiscal policy. Because the moment you give one company a special deal, every other company wants one too. And the competition only escalates. But if you build a great environment for all businesses, you don't have to write checks to anyone. Luna: It's a harder sell politically, though. Cutting a ribbon for a new Amazon office is a lot more photogenic than funding a new teacher training program. Lucas: That's the dilemma. The incentives are visible and immediate. The payoff from better schools or roads takes years. But the fiscal evidence is pretty clear: the megadeals don't deliver the returns they promise. And with state budgets already stretched — especially after the pandemic — we can't afford to keep burning billions on incentive packages that don't work. Luna: So next time a governor announces a 'landmark' deal with a giant corporation, we should ask: what are we giving up for this? Lucas: That's the question. And the answer might be a lot more than we think.