Evaluating State and Local Business Incentives - Owen Zidar | O-Lab Workshop 2020

So this is joint work that actually just came out in the Journal of Economic Perspectives with Cailin Slattery, who just started at Columbia Business School. And we have some follow-on work that's going to be related to this as well. So the broad motivation is that state and local governments are spending tremendous resources trying to attract and retain firms. And these incentive policies are quite controversial. So some people argue that attracting firms like Foxconn and Amazon to local areas is really key for local economic growth and development, but others question how much we've been spending on these policies and how effective they really are. Trying to inform this debate is pretty difficult because there are a number of empirical and conceptual limitations. So empirically, there's just not that much data on how prevalent business location subsidies are, how big they are, who gets them. And there's also not a lot of transparency on the selection process that determines where firms ultimately go and what bids existed. And then a third problem is the fundamental problem of causal inference. We don't know what would have happened in the absence of these subsidies. And so What we're going to do in this paper is try to build on some, some new data that Cailin put together in her job market paper, which is an excellent paper that you should, you should read, and we're going to try to make progress on these 3 steps. And so the, the 3 things that we're going to try to do is characterize incentive policies using data that Cailin put together, describe the selection process that determines which places and which firms give and receive incentives, And then try to shed a little bit of light on what happens economically after some of these incentive deals occur. To summarize our findings, in terms of what the policy actually is, states are spending a wide range, something between $5 and over $200 per capita on incentives for firms in a range of forms. Some of those are direct firm subsidies. Some of those are tax expenditures. And overall, we come up with an estimate of state and local governments spending at least $30 billion on business location incentives. That's a little bit lower than some other estimates, and we'll compare them and talk about some of the methodological differences for coming up with these numbers. And in terms of the average discretionary subsidies, so that's a firm-specific subsidy, It amounts to $180 million over a 10-year period, and it's roughly for about 1,500 jobs. So you can think of a stylized example of an auto plant. They're gonna pay a subsidy and get one large auto plant to a location. And in terms of selection, the firms that get these are often quite large and quite profitable firms. We see more prevalence in manufacturing, especially in auto and aerospace. As well as tech and finance, but there are a range of industries that have received these subsidies. Firms tend to accept deals from places that are richer, larger, and more urban. And perhaps somewhat in response, poor places tend to be more generous to try to attract firms to poorer areas. In terms of impacts, what we do is we follow The methodology of Greenstone, Hornback, and Moretti, and we use that in more recent data for a broader set of firms. And what we do is we compare places that get to it, that successfully attract a firm, to their runner-up locations, and we do see some evidence that there are direct employment gains from these incentive deals. Now, I should say, we're— the experiment that we're doing is a little bit different than some of the prior work that looked at the effects of investment per se. It's not necessarily the case that the effects when a firm does a large investment is the same as when a place does a large incentive deal. In the former, there's actually investment. In the latter, there's not necessarily big real investment. Um, so what we find is kind of more mixed evidence on spillovers. We don't see a lot of effects beyond these kind of direct employment gains, um, and I'll go through some of that evidence in the third part of the presentation. Um, now, these things could be completely consistent, um, because we're kind of answering slightly different questions. What happens when there's an incentive versus what happens when a firm, um, makes a large investment? All right, so what we're gonna do, um, is we're gonna start by I'm trying to have a conceptual framework to think through some of these policies, give a specific example of how this works from a Volkswagen deal, and, and then provide some new facts on these incentive deals, and then talk about the consequences and policy implications. So I think it's useful to start by just thinking about what state and local policymakers say they're trying to achieve. And then think about the different tools that they have to achieve those objectives and what the costs and benefits and trade-offs are of these different objectives. So if you look at some of the statements of local economic development policymakers, what they'll say is, you know, they're trying to recruit and attract new business and industry. The state senator from Maryland said it's about jobs, jobs, jobs. So one way that we like to think about it— so you think about kind of a social welfare function, you consider how different policies like firm-specific subsidies or other types of incentives affect different agents in a state and local area. So the 3 agents that, that we like to think about are workers, capital owners, and politicians. So workers care about, you know, having good employment opportunities, higher after-tax real wages, good government services. Capital owners care about their take-home pay after you account for tax and incentives. They care about product demand and productivity and factor prices, which determine their take-home pay. And politicians benefit from things that affect their reelection odds, campaign contributions, or opportunities to provide pork. And there are 3 main ways that we consider in the paper, um, you know, affecting those, those payoffs. So one is to give every firm in the local area a lower corporate rate. A second is to narrow the corporate base. So this would be something like if you do R&D incentives, you're not going to have to include that in the tax base, and that will effectively lower your tax bill. And then the third that we'll consider is offering a specific firm a tax incentive. And, you know, what's pretty striking is if you look empirically, states take quite different approaches in terms of how much they use these, these different methods. So if you look for the first instrument, for corporate taxes, some states have high corporate taxes. So California nearly has a 10% rate, whereas neighboring Nevada has 0% rate, and they're collecting widely different amounts of corporate tax revenue per person. Secondly, in terms of base incentives, things like tax credits or economic development spending, that also varies quite a bit. And finally, for the third instrument, in terms of the number of subsidies, some states are way more active in terms of subsidy provision compared to others. And the, you know, simple measures of generosity like subsidies per job vary by, you know, an order of magnitude. And on average, it's about $45,000, but in some places it's much less generous. One thing to keep in mind is that these things are typically quite correlated. And so Places that have high state corporate tax rates like California tend to offset that somewhat by spending more on kind of base narrowing provisions. Whereas other places like Virginia have a lower rate but a broader base. It's kind of interesting in the US at the federal level, part of the debate before the Tax Cuts and Jobs Act about corporate tax policy was basically, should we be more like Virginia, where we have a low rate and a broad base, or something more like California, where we have a high statutory rate? And I think there's a lot more work to be done using the labs of democracy to think about the trade-offs between broad base and narrow rate in terms of business tax policy. OK, so what are the costs and benefits of these different provisions? Or different instruments. So for firm-specific incentives, the way that we think about it is that what really happens is for the directly incentivized firm, you know, the welfare gain depends a bit on what happens to local labor demand and wages. But there's also a potential cost for workers to the extent that they're going to have to pay for this through higher taxes or lower public good service provision. And there's also some congestion worries. So when you saw kind of reports of Amazon going to Queens, one of the things that people really cared about was what that might do to local, local prices and in terms of congestion. For capital owners, you know, there's a direct gain for the firm that gets the incentive. And then for other firm owners, it depends a bit on what happens to their productivity. versus congestion. So the extent that there's big spillovers in terms of productivity or product demand, um, that can be quite helpful, but it could bid up wages or land costs or taxes. And then quite relevantly, politicians might benefit quite a bit from firm-specific incentives because there's a salient plant that they helped try to keep And there's goodwill from the recipient firm. And that can differ a bit from what you would normally get for these different groups if you just lowered corporate taxes for everyone across the board. So for lowering corporate taxes, workers could benefit if it increases labor demand and wages. But, you know, there are some cases like in Kansas where it seems like public service provision fell enough such that, you know, you can overdo it in terms of lowering corporate rates. In terms of capital owners, lowering corporate rates is much less targeted. You know, all firms who are C corporations in the state will benefit directly from getting lower corporate rates. And there are some potential impacts on public service provision, productivity, and factor demand. And then politicians could still generate goodwill. from incumbent firms, but it's a little bit less salient who is actually benefiting. In terms of narrowing the corporate base, it's a little bit in between the two, the two cases. So in terms of trade-offs and why we might want to do firm-specific incentives, there are some good arguments. Ed Glaeser has a nice paper that kind of highlights some of these. One is that, you know, it might make sense to try to attract high-benefit firms. There might be some firms that could broaden the industrial mix of a location, or, you know, as Daron Acemoglu and David Autor have pointed out, sometimes it's very valuable to retain a firm because there's a lot of product market and labor linkages that, I mean, if you lose a specific firm, it could take a long time to rebuild relationship-specific capital. In addition, you can also contract specifically with these firms on investment and hiring goals, whereas you can't do that by a broad-based cut in the corporate rate. So you can say, we'll give you this incentive in exchange for a certain amount of investment. And then from a kind of revenue efficiency point of view, it allows states to do some price discrimination and potentially target mobile firms more efficiently. And so those are some of the main arguments for having firm-specific tax incentives. However, in practice, it might be quite hard to pick winners effectively. You know, the political process is involved, and so both identifying and then directing funds to high economic benefit firms might not happen. It's quite difficult to estimate agglomeration effects in practice, and so the idea that this is happening effectively through the political system seems to involve a number of assumptions. Second, it's quite hard to know if firms are actually marginal in their location decisions, and, and so there might be, you know, some of the benefit of only targeting marginal firms might actually be harder to do in practice. One key problem is that often there's a lack of transparency, and when you give too much discretion, And this can result in pretty substantial lobbying and corruption. We've seen examples of this in New Jersey quite recently. And then, as Cecile will point out in a little bit, and this is going to be one of the themes we're going to try to emphasize, is that when you think about the arguments for place-based policies, and in particular these business incentives as a place-based policy, there's— it's good to keep in mind equity and efficiency reasons. From an equity point of view, most poor places might not be able to afford to compete in terms of providing generous firm-specific incentives. So if you're thinking that the primary— or one of the primary goals is to achieve equity goals because of state balanced budget requirements, that means that this tool is not going to be very effective relative to some other tools. The last point is there are probably costs and benefits from a fiscal point of view from providing these incentives. And so it depends on kind of the current level of incentive provision. So if we start from a level that's pretty high in terms of firm-specific incentives, if we want to give one more dollar firm-specific incentives, we might have to finance that through higher taxes elsewhere. Do you think that the gains from these incentives are somewhat diminishing in terms of returns and the costs of raising taxes grows with the square of the tax rate? There's probably a point where we're doing too much, and it's kind of an open question empirically to try to think about how much subsidy provision would make sense. That's not something we tackle, but I think it's an interesting area for future work. Okay, so that was most of the conceptual discussion. Now let me be a little more concrete to give you a sense of how this works in practice. So there's a good case study of the Volkswagen deal in 2008 in Tennessee. So the arrangement was as follows: Volkswagen agreed to go to Chattanooga, Tennessee, and in exchange for Hiring about 2,000 people and doing $1 billion of investment, it got from state and local governments about $560 million. And so what that entailed were about $200 million of local property tax abatements, some state and local job and investment tax credits, which were another $200 million. They also gave property to Volkswagen and some training. As well as funds for improving kind of local infrastructure, and they also had some tax credits for helping local suppliers. It's somewhat interesting to read the reports of what, what the economic development officials thought would happen as a consequence of this deal. So they estimated that would have about a $100 million increase in annual payroll. Which isn't crazy if you think that 2,000 people getting $50K each, that seems to be somewhat reasonable. In terms of the total number of jobs, you can see that 14,000 is 7 times the number of direct jobs. That could happen, but it's not obvious that that would happen. And so the overall economic benefit really depends on these spillover effects, which are somewhat unclear empirically, and we're going to talk about that. In a little bit. Okay, so, um, what Volkswagen did was, you know, they had to make a location decision, and according to, um, you know, descriptions of people who are making the decision, apparently it actually was quite a close competition. They considered about 100 sites, and then ultimately the runner-up was Huntsville, Alabama, which has somewhat similar tax policy. It's also a right-to-work and slightly higher wages, but broadly somewhat similar. And so one of the things we do is we look at kind of employment at the 3-digit level in the auto sector. So it's NAICS 336, and look at how that evolves before and after the deal in the winning place, Hamilton, Tennessee, and the runner-up place, Huntsville, Alabama. What you can see if you look at the graph on the left is that employment in the 3-digit auto sector went from about 1,000 up to nearly 4,000. And if you just use that diff, that can be somewhat informative of what happened. Or you can do a diff-in-diff by comparing that to the change in Huntsville. The diff-in-diff, it might be a little misleading if you think that 2008 was a bad time because of the recession, but I think this gives you some evidence for— in this case, it seems pretty clear that there are kind of direct employment effects from bringing in the auto plant. Okay, so that, that's kind of one deal. What we want to do next is turn from anecdote to more systematic data, and so I'm going to describe that now to talk to you about what happens in general for these state and local business incentive Deals. Okay, so there's 3 main ways to measure business location incentives. The first is a rules-based approach, which Tim Bartik has done. And so the way this works is you collect rules on different incentive programs and tax rates for each area. And then what he's done is he said, suppose I were building a new manufacturing plant in a given place, given all of these rules, what would my payoffs be? And, and that's one way to characterize these incentives. A second way is an expenditure-based approach, which says, let's read all of the state and local budget documents and see how much they're spending. Now, that's different than the rules-based approach because for a given tax rate, if a lot of people take it up, you're going to spend a a lot more than if not that many people take it up. And so it's P times Q rather than P, but it can still be informative. And then the third approach is a more narrative-based approach, kind of like Romer and Romer style, but for individual incentives, where for large incentives, say like the Amazon one, the Foxconn one, etc., you can read newspaper accounts and other, other stories about what was going on. And assemble a dataset that way. And so what we're going to do is we're going to build on the second and third approach based on Caitlin's work. So she very carefully went through state and local government expenditures and put together a novel dataset based on that, as well as a narrative-based approach that looked at specific deals that were above $5 million. Now, to start, one of the first facts I want to document for you is how big some of these different activities are when you take these different approaches. So if you ask the question, how, how much state and local incentive spending is there in the United States using a rule-based approach? Bartik estimates that there's roughly $45 billion in 2015. You know, this is a massive amount of money. In normal times, it's like on the same order of magnitude as unemployment insurance— not unemployment insurance today, but, you know, during non-recessionary times. Of that $45 billion, a little under $15 billion was attributed to local property tax abatements. So the expenditure-based approach that we have from Kaelin's job market paper only looked at the state level. And so at the state level, she found about $20 billion of expenditures. And so if you think that, based on Bartik's work, that 1/3 of the overall state plus local spending is done at the local level, that means our $20 billion at the state level amounts to about $30 billion in state plus local. Now, we, we think that this is probably a lower bound for some of these incentives. And another estimate I should point out is from the New York Times. So they had a An estimate of $80 billion in 2012. My, my understanding, and Caitlin knows this better, so feel free to follow up with her if you're interested in these programs, is that there's a bit of double counting in some of those numbers because they combined some state tax expenditure reports with specific deals, and those are in both data sources, and they didn't always account for that. And so The $80 billion estimate might be a little bit too high, but there's a lot of uncertainty given how many programs there are and the reporting requirements aren't very good. So what, what we're going to do is I'm going to document about 4 or 5 facts using this data from Kalen on state-level expenditures, as well as a dataset of about 540 deals that happened between 2002 and 2017. The sample procedure for defining the sample is we took all of the Good Jobs First data, which is an organization that tries to track subsidies, and we kept all the ones that were at least $5 million and involved a discretionary program that explicitly mentioned expansion or relocation. So there's some other discretionary programs. that are not included there, but this is kind of most of the— we're trying to attract or retain a firm type of activity. And what Caitlin did is she read news articles about each of them to fill in details about runner-up, subsidy size, how many jobs were promised, etc. Okay, so the first fact from, from these data is that large companies are much more likely to receive incentives. So if you look at The recipients of these deals, they're much more likely to be, you know, large 1,000+ employee establishments than the typical establishments in the United States. If you link the data to publicly available data in Compustat and see how big these firms are, these are quite sizable firms. So in the standard Compustat firm has about $3 billion of market cap. The typical firm that's subsidized here is $76 billion in market cap. So these are kind of big profitable firms among publicly traded ones. And one of the things we do is we look at the most prevalent industries. And, and so what we did, we have some tables showing the top 10 most prevalent industries. And what you see is of the 543 deals, about 10% of them were in automobile manufacturing. A lot were also in aerospace manufacturing, financial activities, and kind of tech, as well as chemical manufacturing. So the characteristics vary a little bit, but for manufacturing deals, you can see the jobs promised are quite sizable. It's, you know, 2,000 to 3,000 people, and the cost per job numbers are hard to compute exactly. So what this is, is just the ratio of the subsidy to the jobs promised. Now, the jobs promised is a little bit different because it could be the case that, you know, these people are hired for 10 years because the subsidy is measured over a 10-year period, but the Employment is, is a one-time thing. And so it's hard to know exactly what the cost per job is. You can take the ratio, but you take these numbers with a little bit of a grain of salt. One way to think about it is you could divide the cost per job by 10 to have an annualized measure. So this is something like in auto manufacturing, this would be like a $10,000 cost per job year estimate. Now, Across these industries, you see pretty different levels of generosity. If you look at basic chemical manufacturing, there's some deals, especially in Louisiana, where the numbers of jobs promised are really quite small, but the subsidy is quite big. So you're getting $1 million per, per job, or, you know, $100K per job year. There's also investment levels that are important. And quite big that vary. And so we provide more of that detail in the paper if you're interested. In terms of the spatial component, it's pretty interesting. If you plot the cost per job measure, so that's the subsidy divided by number of promised jobs on the y-axis, and you relate that to average wages in the local area, what you see is that rich places like San Francisco tend to see, or tend to provide incentives that are much less generous on a cost-per-job basis relative to some other places like Charleston, South Carolina, or Albany, New York. So, poor places tend to spend more per job. The last descriptive fact that I'm going to cover is a regression that tries to characterize when these incentive deals happen. So the, the left-hand side variable is an indicator for per capita incentives increased by 20%, and different right-hand side variables are related to political outcomes, like is it an election year. So the second column shows that, you know, incentive spending is much more likely to increase if it's an election year. The Later rows, kind of 3, 4, and 5— or sorry, 3 and 4 show some economic outcomes like percentage change in EPOP, which does seem to be related to incentive provision. So when, you know, economic conditions are deteriorating, it's more likely to give incentives. But it looks like the political economy considerations, which are shown in the last one, the government can run as incumbent and it's an election year. Has much more predictive power for when incentives are going to increase by a material amount. So this suggests that political economy considerations should be first order when thinking about some of these incentive programs. Alright, so now we can turn to evaluation— what actually happens. So what we're going to do is we're going to Do an event study specification for the 500 deals that we have in this firm-specific dataset that occurred between 2002 and 2012, so that we can look at outcomes 5 years before and 5 years after. And we're going to compare winners versus runners-up, and we're going to control for conditions 10 years before the deal, so particular log population and log average wages 10 years before. So this is what it looks like in terms of overall employment at the county level in the 3-digit industry of the subsidy. So the Volkswagen example was in the auto sector. This is kind of like the generalized version of that graph. What you see in levels is that looks like in the years before these deals that The winners and runners-up were somewhat comparable, and then there's a relatively sharp increase in employment, and it goes up by about 1,500 people, which is where we get that estimate of, you know, there's some evidence of direct job gains when you look at what happens when you compare winners and runners-up. When we look at kind of the residual amount of employment in the 3-digit sector, so this would be like auto employment except for in auto plants, or at the 2-digit level or the 1-digit level, we don't see very much evidence that employment outcomes improve in kind of the broader auto sector. We also provide in the appendix a number of other kind of measures that try to say, well, maybe some of the spillovers vary across geographic region or in different measures. And, you know, we tried a bunch of different things and didn't really seem to find a lot of strong evidence for spillovers in terms of employment. A second outcome, which is quite of interest for the, the urban folks, is if you look at what happens to house price indices. So what it looks like is that in the pre-period, winners tended to have slightly higher house prices. That were not statistically different from, from losers, but again, a little bit higher. And it seems like winning places tended to have a decline in house prices relative to losers on the order of a couple percentage points. And you're 5 years after about 4 percentage points lower than— So it seems like there's some suggestive evidence. It's hard to know if this is a trend. but some suggestive evidence that house prices might have fallen a little bit on average following these deals. Now, that doesn't mean that some of these deals were quite effective and some were not. It's just saying on average, we don't find strong evidence that house prices go up following this sample of deals, which is in a more recent period than some prior work. Now, one of the things we tried to do to understand, you know, how these results compared to what people had done before. So it's important to know that there are kind of different samples and different questions. So for the million-dollar plant paper by Greenstone and Moretti in 2003, there are 82 subsidy deals in the '80s and '90s. And so the conditions were potentially quite different. They were mostly in manufacturing, where now services and tech and oil and gas play you know, a larger role in our sample, only about, you know, less than a quarter of our sample was in manufacturing. And some of the other studies were also looking at establishment-level TFP spillovers or establishment-level effects. So, Bloom et al. kind of updated the data in a more recent period and did find evidence of spillovers at the establishment level. You know, but one thing that's worth pointing out is looking at the effects of a new plant or a new investment is potentially different than what happens as a consequence of one of these subsidies. It could be the case that a firm gets a subsidy and doesn't change any of its behavior. So, the consequences of actual new investment versus the consequences of a firm getting a subsidy can be different. And so that's how we can reconcile our results with prior work that uses different samples or, or different levels of aggregation. So if you're interested in this, we have more discussion of this in the paper. All right, so, you know, so far I've talked mostly about the state and local level, but some of the interesting questions, especially economically, occur when you start thinking about this from a national perspective. So when you want to think about what state and local governments are doing, you know, one natural way to think about it is in terms of the lens of tax competition and a race to the bottom, which has led some people to call for subsidy bans in a similar way that's happened in other, other parts of the world. So one way to think about it is to say, all right, let's start from the simplest possible frictionless benchmark where there's no externalities. everything is at its optimal level, there's no other distortions, government service provision is, is where it should be. And in this case, it's likely that business incentives— introducing business incentives would likely lead to some misallocation and reduce aggregate welfare. You can, you know, think about this from the perspective of Cecile's job market paper or my paper with Eduardo, Juan Carlos, and Pablo. and restud, um, you know, you can see what would happen if there is no other frictions. Now, if you start adding different components, uh, one component to naturally add is to think about externalities. So, um, there can be pretty substantial fiscal externalities and productivity spillovers, uh, which is, um, one of the reasons that, that people have thought, um, some of these policies could make sense. Uh, Ralph Ossa has a very nice paper thinking about about this. And, you know, Enrico has some interesting work thinking about the IT sector and what happens if you, say, locate a given plant in a highly productive place versus a less productive place. It could be in some circumstances and in some industries that the gains you get from locating that plant in a very productive place outweigh the losses um, from a less productive place. Now, um, in Enrico's paper with Pat on the Tennessee Valley Authority, they, they estimated, um, um, uh, an elasticity that they kind of roughly found that those things cancel out, but I think there are some cases where they might not cancel out. We just don't have a lot of evidence, um, kind of in general across most industries and most settings. Um, but, you know, that's a possible consideration to think about when you're thinking about these programs from a national point of view. What I think is empirically first order and something that hasn't gotten as much attention are not just these kind of efficiency consequences about thinking about spillovers and productivity, but also thinking about equity. And, you know, one way to think about why place-based policies might make sense, if you, if you think about You know, we're not starting from a frictionless first-best setup. There's a number of frictions in the labor market. There's a number of pre-existing distortions from the tax system. And so it's quite possible that we're in the world of second best and some intervention could actually improve welfare. Cecile is going to talk about a framework where, you know, she thinks about that much more carefully and formally in a very nice paper. There's also Adrian Bilal's job market paper where he was thinking about labor market frictions in the context of a model of search, and he analyzes some of the consequences of thinking about kind of having jobs in different locations based on labor market conditions, and it seems quite plausible and quite important that the shadow value of a job could be quite unequal across regions. And so we think this should be much more top of mind when thinking about place-based policy. Okay, so where does that leave us? So I think I have a couple thoughts on policy to wrap up. So, you know, if you ask real experts in this area like Tim Bartik, he's of the view that, you know, it can make sense to target tax incentives to hard-hit places, but there should be you know, some provisions that really helped improve the equity benefits of these policies. So, for instance, you can subsidize employers to hire local residents rather than just anyone. And there might be ways to target marginal investments and job creation for industries that have high multipliers in a way that increases transparency and makes political concerns less salient. You know, in general, it's been hard to inform this debate because there's not a lot of transparency. There are some states that are doing a better job on this, but— and there are some reporting requirements that have started to be phased in, but a lot more could be done. Then the last 2 things I want to consider are thinking about what happens if you have harmonization across places. So, one of the things that we found in our restud paper Eduardo, Pablo, and Juan Carlos is that you can get a lot of the gains, the welfare gains from harmonization at just the regional level that you would by making things completely equal across place. And so it's quite possible that some kind of non-compete pacts at the regional level could improve things quite a bit. And we've seen evidence of this. Kansas and Missouri did this. And New York lawmakers are also considering acts of this form. A second approach, which also strikes me as having some promise, is having a larger role at the federal level. You know, so in the, in the EU, there's some scope for this. And I just want to say, in terms of recent police-based policies, you know, they have been less targeted than some of the more successful policies in the past. So federal programs like the Opportunity Zone program, the restrictions on which places were eligible were much less strict than in the Empowerment Zone program in the '90s that Brousseau, Gregory, and Klein studied. If you look at the places that actually got Empowerment Zones, they had really high levels of unemployment and poverty rates relative to the upper— Opportunity Zone residents. So if you're thinking about broadly what types of programs do we want to do for place-based policy, what are the equity and efficiency consequences, I think we can do much better in terms of the equity gains from, you know, having more strict requirements on things like Opportunity Zones. And so there's a lot of room for improvement here. Okay, so in terms— to conclude, the last thing I want to say is for open—

Owen. Yes. Sorry to interrupt for a second.

Yep.

We had some technical difficulties on the Q&A function.

Ah, okay.

So before you conclude, I wanted to direct everybody in the audience, if they want to ask questions, we have now enabled the chat function. So please use the chat function to ask questions. And apologies if the Q&A function is not working. Once they inform us that the Q&A functions start working again, I'll let you know. But for now, use the chat function. Thank you. Owen, it's back to you.

Okay. Yeah, so in the last, you know, minute or so, you know, I think there's a number of open questions and directions for future work. So One is to actually measure the extent to which these state and local business incentives actually trickle down and affect underemployed people and low-income workers. And one of the key policy debates over recent things, like in the Amazon case, it was not clear at all what types of people are going to get these jobs. Are the good jobs going to go to immigrants and high-skilled workers, or are the benefits gonna go to some of the— prior residents, and will it help employment-to-population ratios in the targeted place? A second key question, and Tim Bartnik has done some work on this, and I think there could be a lot more work in trying to think about how effective business location incentives are relative to other approaches. And thinking about, you know, if you're in the perspective of a governor or a mayor, how to think about some of these policy trade-offs relative to, say, taking money and investing in education or other sources. There's clear signs that political economy is first order in the allocation of these programs. And so I think more work to think about political economy, the causes of these incentives. And finally, you know, one of the things we found was that Really large profitable firms are much more likely to get these incentives. And so that can be another force that, that's leading to anti-competitive effects among superstar firms. If it's, you know, not only are you not able to compete, say, on a productivity basis, but if superstar firms are the ones that are much more likely to get these business incentives, that makes it even harder for other firms to enter and do well. And so that's another thing to keep in mind when we're thinking about these firm-specific business incentives.