Good Jobs First and the Business of Keeping Score

For most of the modern history of economic development subsidies, the striking fact was not that deals went bad but that no one kept a systematic ledger of whether they went good. A state would grant a package, issue a press release, and move on. Years later, if a journalist asked whether the jobs materialized, the answer was frequently that no one had checked, because checking was no agency's assigned job. Into that vacuum stepped a small set of watchdog organizations, and one deserves to be named specifically because it built the infrastructure the whole field now relies on: Good Jobs First.

Founded in the late 1990s and led for decades by Greg LeRoy, Good Jobs First is a nonpartisan research center whose insight was procedural rather than ideological. It did not argue, in the first instance, that subsidies were good or bad. It argued that they were public spending and therefore ought to be tracked, disclosed, and evaluated like any other public spending — and then it did the tracking itself, because the governments handing out the money largely would not.

The organization's most consequential product is its Subsidy Tracker, a database that aggregates individual subsidy awards from thousands of programs across federal, state, and local governments into one searchable place. This sounds mundane until you grasp the problem it solved. Before it, subsidy data lived in scattered agency filings, annual reports, and press releases, in incompatible formats, with no common identifier tying a subsidiary back to its corporate parent. A large company could collect benefits under dozens of legal entity names across many jurisdictions, and no one could add them up. Subsidy Tracker's discipline of "parent-subsidiary matching" — linking awards back to the ultimate corporate owner — made it possible, for the first time, to ask how much total public money a given corporation had received. The answers were frequently larger than anyone had assumed.

A companion tool, the Violation Tracker, extends the same logic to corporate misconduct, compiling penalties and settlements so that a subsidy recipient's compliance record sits alongside its subsidy record. The organization has also maintained trackers focused on particular high-subsidy sectors and on especially large deals — the "megadeals" where a single company extracts a subsidy package running into the tens or hundreds of millions. The through-line is always the same: convert scattered, deliberately illegible information into a public ledger that can be searched, summed, and cited.

Why does this matter for clawbacks specifically? Because a clawback is only as good as the measurement that supports it, and measurement is exactly what the watchdogs supply where governments do not. A clawback trigger tied to a job count is meaningless if no one independently records the job count. A proration formula is idle if no one tracks the shortfall. When an outside database makes a company's promised-versus-delivered record visible, it does two things at once: it gives enforcers the evidence to act, and it raises the political cost of not acting. A subsidy failure that lives only in an unread agency file embarrasses no one. The same failure, indexed and searchable in a public database that a reporter can query, becomes a story — and stories move officials who clauses alone do not.

There is a lesson here about how accountability actually arrives in fiscal policy, and it is not flattering to the official process. The scorekeeping that disciplines these deals was, for a generation, supplied from outside the government by people the government did not ask and often did not thank. That is a fragile way to run accountability — it depends on the survival and funding of a handful of nonprofits — but it is the arrangement that produced most of what the public knows about whether subsidies deliver. When you cite a figure about a subsidy's underperformance, you are very often citing, directly or at one remove, a database that a watchdog built because the responsible agency did not. Keep that provenance in mind. It tells you how much of this system's accountability is load-bearing volunteer work, and how much would simply go dark if the scorekeepers stopped keeping score.