Who's Afraid of Sunlight? Explaining Opposition to Transparency in Economic Development

Abstract Firms and governments often negotiate economic development deals, such as tax abatements, with limited transparency, using exceptions to public records laws or other strategies for nondisclosure. In this article we explore the motivations of firms for keeping economic development deals out of the public eye. We explore legal challenges to public records requests for deal-specific, company-specific participation in a state economic development incentive program. By examining applications for participation in a major state economic program, the Texas Enterprise Fund, we find that a company is more likely to challenge a formal public records request if it has renegotiated the terms of the award to reduce its job-creation obligations. We interpret this as companies challenging transparency when they have avoided being penalized for noncompliance by engaging in nonpublic renegotiations. These results provide evidence regarding those conditions that prompt firms to challenge transparency and illustrate some of the limitations of safeguards such as clawbacks (or incentive-recapture provisions) when such reforms aren't coupled with robust transparency mechanisms. We speculate that the main motivation for these challenges is to limit scrutiny of these deals that could lead to backlashes against future economic development agreements.


Introduction
Government transparency can improve public policy as well as public trust in government. Revelations enabled by open records laws have prompted numerous calls for domestic and international government reform. 1 In the United States, the lack of transparency surrounding the economic development activities of states and cities, such as the use of targeted tax incentives to encourage business investment, has been heavily criticized by scholars and activists. 2 Recent efforts by activist NGOs and governments have led to increased transparency in many economic development programs. 3 Yet little is known regarding the effects of these reforms, or about which elements of economic development that governments and firms are attempting to keep out of the public eye.
In this project we take a novel approach to this question by examining the transparency behavior of corporate beneficiaries of economic development programs. These economic development programs, or incentives, are financial support for firms in exchange for promises of capital investment or job creation. These financial benefits for firms can come with political or societal costs. For example, great opposition resulted from the announcement of Amazon's second headquarters in New York City due to the almost $3 billion in incentives offered to the company. Similarly, protest against incentives for Amazon in Virginia and Foxconn in Wisconsin haven't gone unnoticed as well. Additionally, industry trade publications such as Area Development included advice to firms on how to avoid unwanted media scrutiny of these deals. 4 Given the sensitivity of this policy area, we specifically examine firms' responses to public records requests for information on their economic development incentive deals that they received from a state economic development program. We argue that by understanding firm preferences for or against transparency, we can begin to understand the implications of this transparency revolution for policy making as well as the types of activities that firms are attempting to shield from public scrutiny. In short, we uncover what negotiated agreements firms are trying to keep secret and then speculate the reasons why these revelations would be damaging to the firm.
Specifically, in our article, we examine firm opposition to public records requests from the main Texas discretionary economic development program, the Texas Enterprise Fund (TEF). This program selectively awards cash grants to companies relocating or expanding operations in Texas in exchange for job creation. The TEF includes clawback or recapture provisions that allow the state to enforce incentive agreements by delaying payment, canceling contracts, or even forcing repayment of past grants with interest. We made public records requests for TEF contracts and then documented which firms challenged our requests by attempting to withhold or redact parts of their state files.
The state of Texas, and the TEF, is the ideal case to study economic development transparency. First, although the TEF is the largest in the nation, twenty-two states have similar deal-closing funds, and every state and almost all cities provide some type of economic development incentives to firms. 5 Second, the size of the Texas program assures us that there we obtain a larger sample of firms, ranging from energy, technology, health care, and manufacturing to financial services. Third, although every state has different rules governing economic development programs, the lack of transparency and the ability of firms to limit information is commonplace across most states. We discuss the representative and unique elements of this case later in the article. Especially important is an enabling court decision (Boeing v. Paxton) that gives firms (and the government) the ability to limit transparency of economic development programs. We discuss this in more detail.
Our records request, specifically for company-specific applications and contracts, was submitted to the Texas Office of the Governor. Then, the subject companies were notified and given the opportunity to challenge our request. We reviewed these firm challenges to examine why each respective firm challenged our request and what types of information they were attempting to withhold from us. Firms challenged our public records request in 45 out of the 164 total grant contracts.
Although companies were partially successful in limiting our public records requests, we were still able to acquire two valuable sets of documents from the state of Texas. First, the state of Texas released to us a list of all companies that have amended their TEF contracts. These amendment documents are not public and to our knowledge haven't been reported by the media or announced by the state. Second, the state complied with our public records requests by releasing batches of partially redacted amendment and application documents in waves. In total, we have received applications, agreements, and any amendments from 90 of the 164 TEF grants. Some of the documents are partially redacted, but in all cases, we have been able to document how the amendments differ from the original incentive agreement.
What explains a firm's legal challenges to our public records request? We focus on two potential explanations: Firms may be more likely to challenge if they have previously been subject to formal "clawback" penalties for noncompliance, or if they have privately renegotiated their incentive contracts to reduce or delay their job creation commitments to avoid clawbacks.
The state of Texas monitors the performance of the TEF, and has imposed public clawbacks, or grant recapture, for many companies that participate in the program. These formal contract provisions enforce the negotiated terms for job creation and wage levels. The state may also rescind awards for defaulting, along with additional financial penalties. Information detailing such penalties is public, posted directly on the government website in an accessible table format. 4 Harris, 2018. 5 Office of the Governor, 2019.
What isn't public is that a large number of companies have renegotiated their TEF agreements. Renegotiations usually come in the form of committing to fewer jobs created, or changing the hiring schedule, or how headcount should be computed (with some renegotiated deals allowing firms to count employees in subsidiaries that weren't party to original TEF-subsidized project). In many cases, such contract amendments are made right before a company would otherwise be subject to clawback provisions. For example, in one case, an incentive agreement was changed to reduce the number of jobs required just one day before an employment deadline.
Statistical analysis of which companies challenged our public records request suggests that factors such as the size of the incentive or the number of jobs proposed have little predictive power in explaining the challenges. To our surprise, we found that companies that had been subject to clawbacks or recapture in the past were less likely to challenge our request. Conversely, companies that had negotiated secret amendments to their TEF agreements were much more likely to seek closure for their files.
These results suggest to us that the firms that have already been subject to publicly disclosed clawbacks for failing to fulfill the conditions of their agreements had little incentive to block investigations into their applications and agreements. Rather, companies that had made amendments to their agreements, which to our knowledge had never been made public, challenged our public records request. Why might companies want to prevent the public disclosure of their amendments? We point to extant work that finds a strong relationship between negative media coverage and decreasing firm value. 6 If companies fear that news coverage of their amendments will place blame on them for failing to keep their promises, which in turn could reduce the value of their shares, they have a strong incentive to prevent disclosure of the amended agreements.
In the final section of the article we examine a sample of these applications and find a clear pattern regarding amendments. First, a large share of these companies amended their applications to lower their job-creation requirements and thus ease compliance with the conditions of the grant. Our analysis doesn't include any verified external data on the companies' actual jobs, but this pattern is consistent with companies renegotiating their incentive contracts, often also agreeing to smaller grant awards, to avoid public clawbacks. We also found other kinds of application amendments, such as how jobs are defined, and how much of a penalty a company will be obligated to pay in the event of noncompliance with the contract.
Our results also speak to the academic literature on the implications of transparency for firms. Although our study focuses on a single incentive program, the firms involved in this study are large, often multinational firms with the ability to locate across the country. Our study suggests that although some firms renegotiated their contracts to limit the costs of the clawback, many of these amendments were targeted to avoid a publicly disclosed clawback, even if that meant accepting a smaller incentive award. This suggests that the public stigma from being identified as noncomplaint with an incentive award is a motivating factor in renegotiation as well as the motivation for challenging the transparency measures. We suspect that these preferences for avoiding public noncompliance with awards is generalizable across states.
These firm responses are important, but the ability of firms to limit public scrutiny of their participation in government programs is a functions of state and local policies. Our results are troubling for what has been seen as best practices in economic development. Clawback provisions and transparency are considered fundamental safeguards for assuring that taxpayers are protected and that companies are held to accountable in terms of job creation and wages, although there is considerable variation in the implementation of these standards. 7 As noted by Mattera et al. (2012) in their landmark study of clawbacks provisions, many states fail to publicly disclose clawbacks and many states fail to properly enforce facially sound provisions. Our results suggest an additional concern. Companies can not only avoid clawbacks through private renegotiations, they can also use open records appeals to limit the transparency of these programs.
Our research shows that clawbacks and transparency at the start of the process, such as public hearings and the publication of incentive agreements, are not sufficient to assure that companies are living up to their contracts with governments. Governments seem willing to amend these deals when firms fail to meet their obligations, and our research shows that firms can and will use legal provisions to obscure these renegotiations. In this article we don't answer why firms are careful to limit information on contract renegotiations, but we document both a willingness of governments to lower their standards and the firms' strategies of keeping this out of public views. We believe this has important academic and public policy implications.
Economic development programs and firm preferences on transparency Countries, states, and cities around the world use specialized policies to attract investment. This includes the use of investment incentives that can be in the form of tax holidays, tax abatements, cash grants, subsidized loans, or infrastructure for a firm. The drivers for the use of these programs are multifaceted. Proponents of these policies argue that individual incentives can lead to positive economic development spillovers that justify their use, although much of the academic literature is skeptical of this position. 8 Others, such as Li (2006) argue that governments can use these policies to overcome barriers such as the weak rule of law as a means of compensating firms for these other barriers. Rickard (2018) argues that electoral geography shapes these decisions to provide private benefits to individual firms. Jensen (2018) shows that program design can influence the overuse of incentives, even when there is considerable transparency in the program. Jensen and Malesky (2018) argue that the main motivation for the overuse of these programs is for government officials to claim credit for firms that locate in their districts. 9 Despite the considerable attention on the impact of incentives and the potential for rent seeking, surprisingly little has been written on the role of transparency in economic development. Although the empirical evidence has been mixed, increasing the transparency of the policy process has been heralded as one of the keys to reducing government malfeasance and corruption. 10 Studies on corruption have found that increasing transparency can either decrease (Islam, 2006) or increase 11 perceived corruption. Cordis and Warren (2014) argue that transparency can deter corruption by uncovering it, which can have opposing effects on the perception of corruption. Using actual corruption conviction data, Cordis and Warren (2014) find that strong Freedom of Information Act (FOIA) laws in the United States increase the likelihood that corruption will be deterred, uncovered, and/or prosecuted.
Evidence shows that the political environment shapes the creation of transparency laws. In one of the most comprehensive studies on the topic, Hollyer, Rosendorff, and Vreeland (2018) find that democratic regimes are more likely to provide transparency in their economic data, which reinforces democratic institutions and facilitates the attraction of more investment. Wehner and de Renzio (2013) find that democratic states are more likely to have strong fiscal transparency infrastructure, and that the effect is greatest in the presence of high levels of partisan competition. In the US context, Berliner (2014) shows that states enact strong FOIA laws where there is political uncertainty over who will control the government in the future. This political competition finding is one the clearest in the literature. 12 These laws provide transparent institutions that seem to be more effective at providing information than informal requests from government agencies. 13 8 See Jensen and Malesky (2018) for a review. See also Austin et al. (2018) for a review of place-based policies. 9 Jensen, Malesky, and Walsh (2015) find that more directly elected politics offer larger incentives and are less likely to provide oversight of these programs. See Fox (2007) on transparency and policy making. 10 Rose-Ackerman, 1999. Another literature studies the impact of fiscal transparency rules on government performance. See Alt et al. (2002) and Alt et al. (2006) for examples. Bac (2001) argues that transparency can decrease corruption through the detection of illegal activities but it can also increase corruption by providing outsiders information on the key decision makers to target for illicit activities.
Bureaucracies may also selectively provide information based on the attributes of the requestor, including partisanship 14 and gender. 15 Other studies find that the political context, such as the polarization of politics 16 or the perceived behavior of other government agencies toward FOIA requests 17 all impact responses.
Many of these studies examine how variation in bureaucracies or requests affect responses. Little is known about how third parties, specifically corporations, respond to these public record acts. Management scholarship has examined voluntary disclosure in areas such as political spending 18 and financial reporting, 19 but it lacks work focusing on firm responses to disclosure laws. Jensen and Malesky (2018) find that numerous industry associations as well as state economic developers formally protested new rules to improve the transparency of tax incentives. In this study, we explore the different firms' responses to these requests.
Our focus on the preferences of firms on the disclosure of economic development incentives provides a broader window into firm preferences for transparency. Academics have long been critical of these programs, arguing that most economic development incentives provided in the form of cash grants, property tax abatements, or income tax credits are particularly ineffective. 20 A recent metaanalysis of thirty-four studies by Bartik (2018) finds that most incentives are offered to firms that already have plans to relocate or expand in a given location. The studies' findings range from only 2 percent to 25 percent of incentives tipping investment decisions to a given location, with the remaining 98 percent to 75 percent of firms maintaining preexisting plans.
Nationally, economic development incentive programs have come under fire across a number of states, including Texas. The specific program that we are exploring, the TEF, has been the subject of numerous controversies, 21 sparking a heated debate in a recent legislative session regarding whether to defund the program. 22 Other recent controversies include major scandals in New Jersey 23 and Georgia 24 over their economic development programs. These controversies were uncovered by a whistle blowers in New Jersey and a state audit and evaluation in Georgia. In both cases, the lack of transparency of the program was at least partially responsible for this mismanagement.
In some states, firms have a limited ability to withhold incentive information, such as Michigan, which provides full disclosure of firms' contracts. Interestingly, many of these contracts contain amendments, which are often admissions by the company that it cannot fulfill the initial terms of its incentive contract. These contracts are reported in the local press and include negative press stories on amendments that are sought after by companies to provide fewer jobs than initially pledged. 25 Yet this transparency isn't complete, as the Snyder administration recently used a nondisclosure agreement with General Motors to avoid disclosure of its total tax incentives. 26 Thus, Michigan provides us with an example of a state where, due to the publicly available incentive contracts, we can observe the large number of amended contracts. In the next section we examine the flagship Texas incentive program, which is the largest "deal-closing" fund in the country. Unlike Michigan, these incentive contracts are not publicly available and Texas law allows firms to challenge disclosure. We argue that this is the perfect environment for examining how firms respond to transparency requests.

Public information requests and the Texas Enterprise Fund
States, counties, and local governments can provide incentives under different programs, legislation, and state constitutions. There is tremendous variation in these programs, although it is common for many states to provide incentives through forms of tax limitations or abatements or through grants. One of the least constrained of these types of programs are what is called state "deal-closing funds." These programs provide discretionary grants to firms in exchange for capital investment or job creation. Firms sign formal contracts with state agencies and in many cases are subject to formal "clawbacks," or penalties for not meeting these benchmarks. Although the size of these programs varies, to date, twenty-two states have similar deal-closing funding. 27 Texas has several economic development programs, but none are as high profile as the TEF. This deal-closing fund was created in 2003 to target firms considering either Texas or other out-of-state investment locations. According the Office of the Governor (2017, 3), "The fund is used only as a final incentive tool where a single Texas community is competing with another viable out-of-state option." In other words, awards are made on a discretionary basis; they are not "as of right," or automatic based on a company performing an eligible activity. TEF awards can be made only by a unanimous vote of the governor, lieutenant governor, and speaker of the House of Representatives.
The TEF hasn't been without controversy. A scathing audit of the program in 2014 uncovered numerous weaknesses, including companies being awarded grants without filling out formal applications, companies having no job creation obligations for awards, as well as poor oversight and monitoring. 30 Numerous news outlets ran stories on the program and high-profile politicians including the then Texas House Speaker Joe Straus, have criticized the program. 31 In an unrelated case, after this audit, the Supreme Court of Texas issued a ruling that was criticized by transparency experts in the state. In Boeing v. Paxton , the Court ruled that if disclosed records would put governments or companies at a competitive disadvantage, they could legally be withheld. The Boeing ruling has been invoked by numerous firms and government agencies, most famously by the city of McAllen, Texas to shield the details of its contract, including the fee, for an Enrique Iglesias concert. 32 In the context of our study, the Boeing ruling provides a legal avenue for firms receiving economic development incentives to challenge public records requests seeking information on their applications or contracts. As a result, affected parties-including government actors, consultants, and the firms receiving awards-are given ten days from notification by Office of the Governor to formally challenge the public information request.
In the next section we outline our research design of formally submitting public records requests for this single state program. We believe that Texas is an excellent location for a case study to test firm responses to transparency requests for several reasons.
First, as noted earlier, many states have similar deal-closing funds. And even outside of these funds, the competition for Amazon's second headquarters, Amazon HQ2, showed that numerous states were 27 Office of the Governor, 2019. 28 https://gov.texas.gov/uploads/files/business/TEF_Listing_10-11-18.pdf. 29 Typically, the TEF matches, and exceeds local incentives offers. In some cases, firms withdraw from the local incentives programs and continue in the TEF program. willing to enact legislation, such as Maryland's PRIME Act, specifically to attract Amazon. 33 Thus, many states have discretionary funds as well as the ability to craft legislation specific for economic development.
Second, the size of the TEF allows us to obtain a large sample of firms across several industries. The diversity of the Texas economy and the size of the TEF allows for a very wide range of firms to participate in the program. Although these firms are large, often multinational firms, in 2020 and 2021 alone, firm participation in the program included software developers, telecoms, and manufacturing including steel, pet food, plastics, water filtration systems, heavy duty trucks, and foot ware. These firms are all large and, by design, could locate in other states. Although a cross-state public records request could allow for a large sample of firms, the costs and time required to file and wait for responses make this unfeasible. 34 Third, there is considerable evidence that both states and firms conduct economic development deals under limited public scrutiny. In an extensive review of state laws, Edmondson and Davis (2011) finds different strategies across states. Twenty-seven states shield economic development negotiations from public records requests through state statues, while eleven states shield negotiations with prospective investors using the structure and rules of their state economic development programs and commissions.
Although there are few comparative studies of how states and cities limit information on economic development, Amazon HQ2 again provides a window on how communities behave across the country. The signing of nondisclosure agreements between communities and firms, or their consultants, has become commonplace. 35 Although we do not know the content of all the deals, we know that Amazon required finalist cities to sign nondisclosure agreements 36 and that the incentive contract in New York and Virginia contained clauses allowing Amazon to challenge public records requests.

Research design and methods
Our original research project began as a study of economic development in Texas, issuing public records requests for all application materials as well as contracts for the TEF. Given that our original inquiry was not about transparency, our research here is not experimental, nor did we have clear theoretical expectations prior to data collection. Thus, our research design is observational and exploratory, yielding insights into the workings of this program as well as theory regarding firm behavior around transparency.
Our public records request was submitted on 28 November 2017 and it included all applications as well as the formal contracts for every applicant to the program since inception. 37 Application materials include a formal application, two years of financial statements, as well as other materials. Contracts include the formal contract between the state and the company as well as any amendments.
Third parties were given ten days to respond to this request. They presented legal challenges and the Office of the Attorney General issued a letter on 28 March 2018 summarizing their comments. A total of forty-five third parties legally challenged our request including forty-four companies receiving TEF awards. 38 Many of these companies cited the Boeing ruling in their complaints. In the Office of Attorney General's summary (28 March 2018, 3): "ADP, Allstate, Apple, BASF, Charles Schwab, Chevron, CITGO, Comerica, Cordish, eBay, E&Y, Fred's, Fritz, GM, GSF, Hulu, Reuters, USAA, and Westlake Chemical each state they have competitors and the release of their information at issue 33 Cox, 2018. 34 MuckRock has been documenting public requires requests since 2018. In many cases, staff and journalists have submitted two requests per month since 2018 to obtain the formal offers for Amazon's proposed second headquarters. My state and local agencies failed to even answer these requests. https://www.muckrock.com/project/america-bids-on-amazon-175/. 35 Byrne and Ziance, 2020. 36 Calvert, 2018. 37 The Office of the Governor generated a cost estimate of $1,207.60 which was deposited on 19 December 2017. 38 One consulting company involved in economic development impact analysis for applications also responded. would give their competitors an advantage." Some of these firms, along with others, also used arguments on trade secrets to shield parts of the application, including CED, Corrigan, Cordish, Dow, HMS, Kubota, Lockheed-Martin, Riseever, Toyota, and Visa. Many of these same firms, including Apple, Cordish, Fred's, GATX, GGNSC, Payless, and Risever also raised legal exceptions for economic development activities through Section 552.131 of the Texas Government Code. In addition to these challenges, firms included other additional challenges, such as USBC's claim that notes in the margins of the application constitute attorney-client privilege.
We are careful in our interpretation of these challenges given that many of these legal challenges were both broad and initiated by hired legal counsel. For example, Apple's public records challenge, from law firm DLA Piper, was one of the broadest in the group, challenging almost every aspect of the application, including the already public information in the document, using numerous cases as well as exceptions for economic development. It is unclear if this especially aggressive legal challenge was at the direction of the company, the firm, or a particularly ambitious associate at the firm. We simply assume that the challenge, and not the specific details, are done at the request of the company.
As we note in the following text, these legal challenges were partially successful; several the application and contract documents that we received through our public records request included redacted sections. Despite these partially successful legal challenges, this public records request yielded three pieces of data that we discuss in more detail. First, the state of Texas provided a complete list of all companies participating in the programs and the companies that have amended contracts to their agreements. This list doesn't provide details on the content of the amendments, but it allows us to determine which companies made changes to original agreements. Second, the Office of the Governor released redacted application and agreements. We have received a total of forty-eight out of fifty-six amended contracts and can document the changes from the original contract. The missing amendments, listed in appendix B, largely consisted of several agreements that never formally submitted applications to the program or didn't have detailed contracts with which to begin. This was noted by a 2014 state audit and was the subject of several critical media stories about mismanagement of the TEF under then-governor Perry. 39 Third, we have the complete list of companies that challenged our public records request. Thus, we have a dataset that includes all company participants in the program, we know which companies challenged our public records request, and for ninety of the grant contracts we have the contents of the firm's applications, agreements, and any amendments.
For our purposes, we do not focus on the legal arguments made by the third parties in their challenges, although our robust tests and empirical analysis do allow us to exclude certain types of legal challenges. Our main dependent variable for our project is whether a third party challenged our public records requests. Focusing only on the projects that have received funding (not new applicants that have yet to receive a first disbursement), we assigned an independent variable denoting the total awards, and a dependent variable signifying the total legal challenges to our request (see appendix A).
We examine two main explanatory causes. First, TEF has a formal process of allocating incentives after certain job-related benchmarks are met. Clawbacks (or recapture) provisions allow firms that are found to be noncompliant with the terms of their incentive award to be punished. These clawbacks have become best practice in economic development 40 although there are concerns about the willingness of politicians to enforce these sanctions and doubts as to whether such measures improve the performance of these programs. 41 Because clawback events are publicly disclosed on the TEF program website, we code firms that paid a clawback or a similar repayment with a 1, and 0 for those with no repayments. A total of sixty-one firms in our dataset paid clawbacks in the period from 2003 to 2017. We code Clawback Dummy as 1 for any firms that had paid a formal clawback at the time of our public records request (November 2017 Our second explanation pertains to what we believe is the first discovery of amended contracts in TEF data. Numerous project agreements were amended after the initial award, in many cases lowering the number of jobs necessary to fulfill the TEF obligations, changing the definition of a job (sometimes even allowing project headcount to include subsidiaries that weren't included in the original application), or changing the job-creation schedule in its entirety. In most cases, these changes led to both fewer jobs and smaller TEF grants. To our knowledge, these amended agreements have heretofore not been public documents.
Our study codes amendments in two ways. First, we requested a list of all companies that received amendments as a follow up to our public records request. 42 We proceeded to Code Amended Contract as 1 for all such companies.
After receiving this list from the governor's office, we checked this list against a sample of contracts and did notice a small number of discrepancies between the governor's office list and our own documented amendments. We realized that some contracts may have been amended before their projects began, such as a contract with SpaceX, and thus were not listed as having been amended on the governor's list. Thus, as a second coding, we simply coded all firms that had undisbursed funds at the time of our public records request with a 1. The correlation between these two measures is 0.82.
Following the advice of Lenz and Sahn (2020), we utilized probit models with no control variables denoting the firm's legal challenge as the dependent variable. We present four probit models in table 1; the coefficient estimates are marginal effects at the mean (MEMs). Our first model includes only our Clawback Dummy and dummy for an Amended Contract. Across all four models we find a consistent sign and size of the coefficient on clawbacks, although the variable becomes statistically insignificant when we include a control variable for the size of the investment.
To our surprise, we find that firms that have been subjected to clawbacks were 12-15 percent less likely to have challenged our public records request. One interpretation of this result is that these

42
The Office of the Governor has been releasing batches of contracts to us through this public record process. After receiving numerous amended contracts, we submitted a follow-up request for a list of all amended contracts. The Office of the Governor provided us with this list. Thus, we have the full list of firms who amended their contracts, but we do not have access to the content of all the amended contracts. companies have already been revealed publicly for not complying with the agreement and faced a financial penalty, leaving less incentive to shield their contracts from public records request.
Our main finding is that amended contracts, regardless of the amendment process, have a significant impact on whether firms challenged our public records requests. Firms that have amended their contracts are between 15 percent and 24 percent more likely to have challenged our public records request. Our results are consistent to additional robustness test including dummy variables in models 5 and 6.
In table 2 we include additional robustness tests. One important test is how public firms, as opposed to privately held firms, respond to these public records request. A second test includes a measure of the dollars per job created. We use this measure as a proxy for bargaining power of firms. Firms with greater leverage, such as the ability to locate elsewhere, invest in important parts of the state, or have better political connections could secure more lucrative incentive deals. For our purposes we do not theorize why some firms achieve higher dollars per job incentives, but this bargaining power could make it more likely the firms receive amendments when they fail to meet incentive performance criteria. Our results show that our results remain largely consistent with table 1, although some models are no longer statistically significant after adding these controls.
Our results provide suggestive evidence that firms that have amended their incentive contracts are more likely to challenge public records request for information on their projects. Given the small sample size we are careful in our interpretations of these results. Yet, this is consistent with firms blocking records requests to hide details of amended contracts.
Case studies: Company-amended contracts In this section we summarize all the cases of amended contracts in our dataset. One reason to dig deeper into these cases is to provide a descriptive account of what types of amendments are given by governments. We stress that we do not know the full range of amendments that were demanded by firms, but our observations provide us the range of amendments that state of Texas was willing to make for firms. None of these amendments, to our knowledge, were ever made public by the state or even by journalists covering economic development.
A second important element of this case study is to examine which types of amendments were challenged by firms. We know the full universe of amendments that were granted by the state, and our case studies can help us identify which were the types of amendments were granted to firms that challenged our public records requests. We begin by exploring these cases in light of our empirical results in the previous section.
The results of our regression analysis indicate that firms that amended their original grant contracts were more likely to challenge our public records request, and firms that paid clawbacks were less likely to challenge. One possible explanation for this result is that firm challenges are not driven simply by whether the firm underperformed/overpromised, but instead by whether its underperformance was public information. 43 It could be that amending firms are pulling a bait-and-switch by publicly committing to create several jobs, and then privately walking back their commitments afterward by amending their contracts. Amending firms would then have an incentive to keep their amendments private through challenging our records request, as a public release could hurt the firms' reputations.
This explanation accords with recent studies in the fields of corporate finance and accounting that find that media coverage is a key predictor of whether socially undesirable firm behaviors have a negative effect on firm value. 44 The logic is that media coverage creates a credible link between the offending firm and the negative outcome, which is observed by both investors and consumers. If firms' private TEF amendments are made public, they may receive negative coverage for two distinct reasons. First, they may receive backlash if their amendment reduces their original investment/job creation commitments or delays the timeline for investment and job creation. Second, they may receive negative coverage simply because they renegotiated their agreement in a nontransparent manner; news that a company privately changed its plans for using taxpayer funds is likely to result in reputational damage.
Firms with privately amended contracts therefore had an incentive to challenge our records requests: To mitigate the risk that the release of the documents would create negative press, resulting in large reputational and potentially financial costs, they rationally chose to incur the relatively lower costs necessary to file legal challenges. However, the returns to challenging the request are lower if the firm has already repaid part of its grant using clawbacks, as these repayments indicate to the public that the firm has failed to meet its commitments. Firms who have already paid clawbacks have relatively less to lose from the release of their amendment documents, as the scale (if not the exact nature) of their noncompliance is already public information, which may explain why they are less likely to challenge our requests.
To explore the plausibility of this explanation, in this section we analyze the content of the forty-eight amended contracts that we were able to access through our public records request. While we do have the complete list of TEF firms that have amended their grant contracts, we do not have the complete set of all fifty-six amendment documents; some firms were successful in blocking our document requests through a legal challenge, while other missing documents may not have been included due to human error or lack of records. 45 However, we do have more than 85 percent of the amendment documents, and the broad variation in the industry and size of the firms whose documents we have received provides reassurance that our sample is representative of all amendments.
To reiterate, the "bait-and-switch" explanation requires that firms are using amendments to make changes to their original contracts that they want to keep hidden from the public. We argue that the content of the amendment documents supports this interpretation; the majority of the amendments serve to reduce or delay the job creation requirements that firms committed to under the terms of their original grant contracts. We identify four nonmutually exclusive categories of amendment and discuss each in turn. 43 To be clear, this analysis is exploratory and not confirmatory. 44 Ahmad et al., 2016;Carberry et al., 2018. 45 Because they were not available at the time of our original public records request, we restrict our analysis to the fifty-six amendments signed prior to 2018. Information on the amendments that we did not receive can be found in appendix B.
The four primary types of amendment are: (1) technical changes, which cover issues unrelated to contract performance (change in firm name, etc.); (2) changes in job-creation schedule, meaning an extension to the deadline upon which a firm is required to have fulfilled its job target; (3) changes to a job target, meaning a reduction in the total number of jobs the firm is required to create; and (4) broadening of the contractual definition of a "created position," which typically means that the firm becomes able to include jobs created at nonproject subsidiaries and affiliates toward its job obligations. Figure 1 presents three pieces of information about each of the amendments to which we have access. First, each row represents one of the four amendment categories identified earlier. 46 The amendments are evenly spread between the four categories, though technical amendments were least common and job definition expansion amendments were most common. Second, each column indicates whether the firm challenged our records request. Note that the proportion of challengers is highest for the job definition expansion category, and lowest for the job schedule and job target amendments. Third, bolded text indicates that the firm publicly repaid its grant contract in part or in full through the clawback mechanism. Note that, among amending firms, the proportion of nonchallengers who paid clawbacks is greater than the proportion of challengers who paid clawbacks. We discuss the significance of this fact in greater detail later in this article.
Of the four primary amendment types, technical change amendments result in the least substantive alternation to the original contract. For example, in 2016 food manufacturer Cerealto amended its 2014 agreement contract to reflect its name change from Siro Group USA to Cerealto Group USA. This simply changes the name on the contracts but doesn't result in any other substantive changes. Because the majority of these amendments do not reduce the amending firm's job creation commitments, it is unsurprising that we see few challenges from firms in this category; while still private, these amendments do not contain information that would damage the firm's reputation if made public.
Dow Chemical is the exception, as it has made technical amendments on two separate contracts and challenged our records request. However, unlike the other technical amendments, Dow's amendments lowered its job "threshold" (the number of preagreement employment positions that the firm maintains in Texas) to reflect the firm's decision to transfer several its employees to a non-Dow company. Changing the job threshold does not change the job target, job schedule, or definition of a created

46
Some firms combine multiple amendment types into one amendment document; for example, Zah Group amended its 2010 agreement to both delay its job schedule by one year and to allow positions created with its affiliate (Klein Tools) to count toward its job target. For this reason, some firms appear in the table multiple times. employment position. However, the amendment documents reveal that Dow is transferring a larger number of jobs out of the company than the number of new jobs that it has promised to create under the grant agreement; this is information that Dow would likely prefer to keep hidden from the citizens whose taxes funded the grant. Dow's amendments, which allowed the firm to privately move jobs out of Texas while its commitment to create new jobs remained public, are consistent with the bait-and-switch explanation.
The second TEF amendment category involves reductions of the job target, the total number of jobs that the firm has committed to create. These amendments arguably result in the largest changes to the original agreement, as they directly reduce the size of the primary deliverable that the firm is required to produce in return for its grant funding. However, the changes are not one-sided; amendments to the total job creation requirement are paired with amendments to the size of the TEF grant. For example, in 2008 Lockheed Martin amended its 2007 agreement to reduce its job target from 800 to 550 new employment positions, a reduction of 31 percent. The amendment also reduced the size of the grant from $5.48 million to $4 million, a reduction of nearly equal scope (∼30 percent).
The third major amendment type involves changes to the job schedule. This allows firms to alter their yearly job creation targets, or to push back their final deadline. For example, Martifer-Hirschfield Energy's 2009 amendment to its 2008 agreement pushed back its job target of ten new jobs to April 2009. The Martifer-Hirschfield amendment is notable because it was signed on 31 January 2009, the same day that its first annual compliance verification (a report on the number and type of newly created positions) was due. Continental Automotive's 2013 amendment to its 2012 agreement left the original annual job targets unchanged, but shifted each of them two years into the future, changing the job creation deadline from December 2016 to December 2018. Job schedule amendments thus allow firms to fine-tune their plans for job creation (as in the case of Martifer-Hirschfield) or make larger changes to the project completion date (as in the case of Continental).
As figure 1 shows, only two of the eleven firms that amended their job target also challenged our records request, and only one out the thirteen firms that amended their job schedule challenged our request. At first, this may seem to be inconsistent with the bait-and-switch explanation-shouldn't the firms who most directly reduced or delayed their job creation commitments have the greatest incentive to keep this information private? It is important to note that amendments and clawbacks are not mutually exclusive. Many companies, most likely struggling to comply with the agreement, first renegotiate their contracts. Some of these companies, such as SpaceX, fail to deliver promised jobs even after the amendment and are subject to a clawback.
Our case studies reveal that firms who directly reduced or delayed their job creation requirements were also the most likely to repay part of their grant award; nine out of the eleven firms who amended their job target paid clawbacks, and eight of the thirteen firms who amended their job schedule paid clawbacks. Recall that clawbacks, unlike amendments, are made public on the Office of the Governor's website. Because the majority of firms in the job target/job schedule amendment categories have already had their grant reduction made public, they have less incentive to keep the details of the grant reduction private by challenging our records request.
Among the forty-eight amendment documents to which we have access, the most common amendment type is the expansion of the contractual definition of a created job. These amendments allow firms to utilize their subsidiaries and affiliates toward their total job target, effectively treating indirect job creation (e.g., new positions created by an affiliated transportation company as a result of increased output from the new production facility) as equivalent to direct job creation (workers hired for the specific purpose of operating the facility for which the firm received the TEF grant) for their development project.
As an example, in 2007 Comerica received a $3.5 million TEF grant to relocate their corporate headquarters to Dallas to create 200 jobs. In 2012, Comerica negotiated an amendment to its original agreement that expanded the range of jobs that could be counted toward its original job creation target. First, the amendment allowed jobs created with subsidiaries, Comerica Bank and Comerica Management Company, to count toward its job target. Second, the amendment allowed Comerica to count fifteen of its executive officers (including the CEO) as newly created jobs provided that they relocated to Dallas or Houston. As an added benefit, executive officers would clearly have an impact on the average wages of the new employees, making it easier to meet wage requirements.
Amendments that expand the definition of a created job alter the agreement more substantively than technical agreements, but they do so less directly than reductions to the job target or delays of the job creation deadline. This may explain the fact that, as figure 1 shows, firms with amendments in this category were most likely to challenge our request. Allowing jobs created with nonproject affiliates and subsidiaries not only makes the job target easier to meet but also changes the composition of the created jobs from what was originally promised; unlike technical amendments, firms certainly want to keep these amendments private. However, unlike direct reductions to the target or changes to the deadline, firms making these amendments were less likely to need to repay part of their grant through the clawback mechanism. Note that of the seven firms in this amendment category who challenged our request, only three paid clawbacks. This finding is well-explained by the bait-and-switch logic. Firms who expanded the definition of a created job privately rolled back their original commitments; unlike the other amendment types, many of these amendments were not publicized through the payment of clawbacks; a relatively large proportion of these firms challenged our records request.
In summary, these amendments suggest that firms have been able to renegotiate their TEF agreements in ways that are consistent with the "bait-and-switch" logic. Amending firms publicly commit to create a certain number of jobs in a certain time frame, and then privately dial back these commitments through amendment; amending firms who paid clawbacks were less likely to challenge our records request, as clawback payment already serves as a public signal that the firm's original bargain has changed.
As we noted previously, Texas isn't the only state to technically allow for amendments in their contracts. Major incentive programs in California and Michigan both list amendments to their incentive offers. The Michigan Economic Development Corporation (2018) provides a summary of its amendments for fiscal year 2017 in its annual report. These amendments look similar to Texas with a large number of companies making amendments to their contracts. 47 This includes changing the definition of a job (Hanson Solutions LLC, OPS Solutions LLC,), delaying their job creation deadlines (Cosworth LLC, Pro Services, Suvina LLC) and in some cases fully reducing the number of jobs required (Flow-Rite, Micro Industries). Many firms included more than one change in their amendments (Bowers Manufacturing Company, Carhartt Inc.). Thus, we contend that this process of amendments occurs in other states and our research design provides a generalizable finding on the firm preferences on the disclosure of this information.

Conclusion
In this project we harness firm legal challenges to public records requests as data. Our project examines firm performance in conjunction with receiving incentives from the flagship Texas economic development fund, the TEF. The program provides cash grants to firms in exchange for meeting job creation targets that include both the quantity of jobs and the average wages of these jobs.
In our request for the original applications and the formal agreements, firms used an economic development exception in Texas public records laws to formally challenge our request in 45 out of the 164 total contracts. In our statistical analysis we show that the firms that were most likely to challenge our requests were firms that had privately amended their grant contracts. Our select case studies illustrate the various types of amendments, but the most common amendment lessens the burdens for job creation through changing the number of jobs required, allowing firms to include employees not originally part of the agreement, or lengthening the amount of time companies have to create promised jobs.
We believe this article has two important implications for economic development. First, numerous amended contracts suggest the ineffectiveness of formal clawback mechanisms if they are not coupled with strong transparency provisions. Allowing companies to renegotiate contracts outside of the public eye violates the very sprit of adding performance requirements and performance provisions. This is an 47 The state lists a total of thirty-nine amendments in fiscal year 2017. During this same period, 107 new incentive approvals were made. Unlike Texas, some of these firms include amendments to their local incentive requirement in the state amendments.
important public policy point that we cannot stress enough. The state of Texas and, as we note earlier, other states have been willing to amend contracts with firms. These amendments, in almost all cases, lower requirements for firms, often in the context of job creation. Although states vary in their transparency of their economic development programs, we believe documenting the willing of a state, often one day before a company would be seen as breaking a contract, to renegotiate is an important finding.
Second, we provide evidence consistent with firms using exceptions to public records requests to hide noncompliance with economic development agreements. While our project isn't designed to test why firms attempt to hide noncompliance (or from whom), we suggest that they may be attempting to prevent media outlets from uncovering the private renegotiations and running negative stories about them. Recent studies have established a strong negative relationship between critical news coverage and firm value, 48 creating powerful incentives for firms to prevent journalists from accessing potentially contentious information (such as private contract renegotiations). Future work should examine the relationship between firms' economic development performance, news coverage, and firm value in greater detail.
We note that there is no difference in our empirical analysis between publicly held firms and private firms. We also note that although a few firms challenged basic information such as financial statements, a strong predictor of legal challenges were the existence of amended contracts. Thus, firms could challenge disclose of documents for any number of reasons, but the secret amendment of contracts is a strong predictor. Whether firms want to keep these contracts out of the public eye, or the fact that they were given special treatment by the government is unclear to us.
What also remains unexplained in our project is the ability of some firms to completely avoid public clawbacks while others firms pay penalties. Our conjecture is that traditional bargaining models of firm government relations 49 explains the state's willingness to renegotiate contracts with an investor. This could range from employment in important areas of the state, the promise of additional investment in the future, or the political connections of the firm receiving incentives. Although our empirical analysis controlled for the incentives per job, a proxy for bargaining leverage at the time of the initial agreement.
It is important to note that other political factors could also shape the willingness of the governor's office to provide more favorable terms to some investors. This could be due to strong relationships with firms through direct or indirect campaign contributions. Equally important, these investments are often touted as major "successes" at ribbon-cutting ceremonies and are announced through formal press releases from the governor's office. Some of these investments may be particularly high profile, and thus the perceived success is important to elected politicians.
Our findings point to clear public policy reforms. First, firms (and governments) are often creative in how they avoid public scrutiny of economic development deals. Even for governments that reveal details of negotiated incentive agreements, these agreements can be amended. Transparency requires not only the original deal but also any amendments or changes to the original deal. Second, although clawbacks are touted as holding companies accountable for their promises and protecting the taxpayers from problem economic development deals, our research shows that the amending of contracts, protecting firms from clawbacks is widespread. This not only calls into question the effectiveness of clawbacks but also challenges the claims that governments want to hold firms accountable. The failure of an economic development agreement not only makes a firm look bad, it can open up the government to criticism as well. This points to need to have an independent enforcers of economic development agreements and clear rules on amending contacts.