Concentration is a policy choice. We are going to keep score.

What the public record already shows

Start with the numbers that anyone can check. In 2025, 1,792 of the 3,104 U.S. counties and county equivalents with FDIC-reported branches had a deposit HHI above 2,500, the highly concentrated line in the 2010 federal merger guidelines. The 2023 guidelines lowered that line to 1,800, so every one of these counties clears both. In 2015 the count was 1,670. Exactly one bank organization serves 187 counties; ten years earlier, 143.

The Census Bureau's 2022 Economic Census publishes a four-firm share for 897 detailed industries. In 12 of them, four firms account for at least 90 percent of the industry's sales, value of shipments, or revenue, whichever measure the Census publishes for that sector. In 43, at least 80 percent. Of the 781 industries that can be compared with the 2017 census, 406 became more concentrated.

In fiscal 2025, companies reported 2,006 transactions under the Hart-Scott-Rodino Act. The agencies opened 41 in-depth investigations, which is 2.1% of the 1,944 deals that could have received one. About one eligible deal in 47 gets a hard look.

Our position

None of those figures is a legal finding, and we will keep saying so on every page that carries one. But together they describe a country whose markets have narrowed in measurable ways while the machinery built to notice has been running at a fraction of its capacity. That is a policy choice. It can be made differently.

Antitrust Radar takes the side of enforcement. We think the concentration screens that the government itself publishes should trigger investigations more often than they do, that bank merger review should look at the county where a customer lives rather than the state, and that agencies should publish the market-share tables behind their decisions so the public can check them. We will argue those positions from the data and say plainly when the data cannot carry an argument.

What Antitrust Radar is

Antitrust Radar is a for-profit research center and newsroom. It is not a nonprofit, a university program, or a government body, and it does not claim to be. Pieces carry the organization's byline. The people behind the work are listed on the People page, the methods on the Methodology page, and every source in a ledger with the date we retrieved it.

The work is paid for three ways, and all three are disclosed here before the first dollar arrives: a newsletter, free now with a paid tier to follow; sponsorships and institutional partnerships; and commissioned research and data licensing for newsrooms, firms, and policy groups that need a concentration analysis of a specific place or industry. As of launch the organization has no outside funders. When it does, no sponsor, client, or licensee will see a piece before publication or shape what it says, and the sponsor of a piece will be named on it.

What we will publish

Three things, on a schedule. Concentration Watch is a weekly column that tracks merger review, enforcement records, and the concentration figures that changed, with a running tally. Data stories rank what the datasets show: the counties with one bank, the states where deposit concentration rose fastest, the industries where four firms hold 90 percent. Editorials say what we think should happen and why.

Every piece links to the page that holds its numbers. The county, state, industry, merger, and enforcement pages stay what they are: reference pages with their limits stated, downloadable, reproducible, and free. Analysis is the layer on top of them that has an opinion.

The line we will not cross

A concentration measure is a reason to look, not a verdict. An HHI does not define a market, a four-firm share does not prove market power, and a complaint is not a finding. We will not publish composite scores that blend measures with different definitions, and we will not present a screen as a legal conclusion. When we are wrong, the Corrections desk records the change, with the date and what was changed.

Hold us to it. The sources, methods, and data downloads are open so that anyone, including the companies and agencies we write about, can rerun the numbers.