DOJ Antitrust Cases: The Federal Record of Government Competition Enforcement

When the United States government decides that a company has grown too dominant, that a deal would lessen competition, or that rivals have secretly agreed to fix a price, the case is usually brought by one office: the Antitrust Division of the Department of Justice. Its litigation record is the spine of American competition policy—roughly 920 matters running from United States v. Microsoft through the criminal cartel program to the modern wave of big-tech monopolization suits—each one a civil merger challenge, a monopolization case, or a criminal prosecution, keyed by the matter and the defendants it names. It is the federal record of who the government took to court to keep markets open.

This article covers what the Antitrust Division's case record is and how it is structured; the two foundational statutes—the Sherman Act of 1890 and the Clayton Act of 1914—and the conduct each reaches; the three kinds of matter in the data, civil merger challenges, civil conduct cases, and criminal prosecutions, and how they differ; the distinctive criminal authority that sets DOJ apart from every other antitrust enforcer and the leniency program that drives it; the division of civil labor with the Federal Trade Commission and how the two agencies allocate matters by industry; the Hart-Scott-Rodino premerger review process that sits upstream of the merger cases; the landmark sweep from Microsoft to Google search and ad tech; how the case record joins to the FTC enforcement data and the broader DOJ prosecution record; a Python workflow that pulls cases, classifies them by type, and tallies them by year; and the caveats every analyst must hold before drawing conclusions from a litigation record.

What the dataset is

The Antitrust Division does not regulate by rulemaking the way an agency like EPA does. It enforces the antitrust laws the way a prosecutor and a civil litigator do—by bringing cases. Its public record is therefore a record of matters: the merger it sued to block, the monopolist it sued to restructure, the cartel it indicted. The Division publishes this record through its case filings and the announcements that accompany them on justice.gov, the public-facing record of what the government's competition enforcers actually did. In our database that record is the table doj_atr, with the grain of one row per matter—roughly 920 cases—keyed by the matter name and the defendants it names.

Antitrust matters carry the distinctive caption of federal litigation: United States v.followed by the defendant or defendants, because the United States itself is the plaintiff (or, in a criminal case, the prosecuting sovereign). A single matter may name a single firm, several merging parties, or a long roster of individuals and companies in a cartel prosecution. The columns capture who was sued, by whom, in what court, when, under which statute, and how it came out:

matter_title          -- the case caption, e.g. "United States v. ..."
defendants            -- the firm(s) and/or individual(s) named
case_type             -- merger (civil), civil conduct, or criminal
statute               -- Sherman Act Sec 1 / Sec 2, Clayton Act Sec 7
court                 -- the U.S. district court (or division) where filed
filing_date           -- date the complaint or indictment was filed
industry              -- market / sector the matter concerns
relief_sought         -- block, divest, enjoin, fine, imprisonment
outcome               -- consent decree, litigated, settled, plea, dismissed
disposition_date      -- date of judgment, decree, plea, or dismissal

The case_type and statute columns are the load-bearing fields, because together they tell you what kind of antitrust matter you are looking at and which law it rests on. A merger matter is a civil suit under Clayton Act Section 7 to stop or restructure an acquisition; a civil conduct matter is typically a Sherman Act Section 2 monopolization suit or a Section 1 restraint-of-trade case seeking an injunction; a criminal matter is a Sherman Act Section 1 prosecution of a hard-core cartel, where the remedies on the table are corporate fines and prison terms rather than divestitures or injunctions. The defendants field is the join key to the rest of the enforcement universe—it is how an antitrust matter links to the same party's appearance in the FTC enforcement record or the broader DOJ prosecution data—and the outcome field is what separates a deal the government merely challenged from one it actually blocked, a distinction that, as the caveats section stresses, a litigation record records imperfectly.

The Sherman Act and the Clayton Act

American antitrust law rests on two statutes, and almost every matter in the dataset is brought under one or both. The Sherman Act of 1890 is the foundation. It was Congress's response to the great industrial trusts of the late nineteenth century—the railroad, oil, sugar, and tobacco combinations that consolidated entire industries—and it is written in sweeping, almost constitutional, terms. Section 1 of the Sherman Act prohibits contracts, combinations, and conspiracies in restraint of trade: agreements among competitors or others that unreasonably suppress competition. Section 2 prohibits monopolization, attempted monopolization, and conspiracies to monopolize: the acquisition or maintenance of monopoly power through anticompetitive conduct rather than through superior products, business acumen, or historical accident. The two sections map onto the two great categories of antitrust harm—collusion among rivals (Section 1) and the exclusionary conduct of a dominant firm (Section 2).

The Clayton Act of 1914 was enacted because the Sherman Act, for all its breadth, reached conduct only after the harm had occurred. The Clayton Act is preventive: it reaches practices that tend to lessen competition before they ripen into a full Sherman Act violation. Its most consequential provision for this dataset is Section 7, which prohibits mergers and acquisitions whose effect “may be substantially to lessen competition, or to tend to create a monopoly.” Section 7 is the legal basis for every civil merger challenge the Division brings—the authority to sue to block a deal, or to require divestitures as the price of letting it proceed. The Clayton Act also addresses other specific practices, including certain exclusive dealing and tying arrangements and interlocking directorates, and it created the private treble-damages remedy that fuels the much larger world of private antitrust litigation alongside the government's cases. A crucial feature of the antitrust laws is that, unlike most regulatory statutes, the Sherman Act is also a criminal statute—a Section 1 violation can be prosecuted as a felony—which is the foundation of the Division's distinctive criminal authority discussed below.

Three kinds of matter: mergers, conduct, and crimes

The case record divides cleanly into three families, and understanding the difference between them is the key to reading the data, because they involve different statutes, different standards of proof, and radically different remedies.

Civil merger challenges are the most numerous and the most visible category. When two companies propose to combine and the Division concludes the deal would substantially lessen competition in some market, it files a civil suit under Clayton Act Section 7 to block the merger or to force the parties to divest assets so that competition is preserved. These are forward-looking, predictive cases: the court is asked to judge the probable competitive effect of a transaction that has not yet happened, weighing market definition, market shares and concentration, ease of entry, and the likelihood of anticompetitive effects against claimed efficiencies. Many merger matters never reach a full trial—the parties either abandon the deal once sued or agree to a consent decree that lets a modified version proceed with required divestitures—which is why the outcome field on a merger matter so often reads “consent decree” or “abandoned” rather than “litigated to judgment.”

Civil conduct cases target the behavior of firms that already hold market power. The archetype is a Sherman Act Section 2 monopolization suit, in which the government alleges that a dominant firm has maintained or extended its monopoly through exclusionary conduct—tying, exclusive dealing, predatory practices, or the denial of access to an essential input—rather than by competing on the merits. These are the hardest, longest, and most consequential cases the Division brings: United States v. Microsoft was a Section 2 case, and so are the modern suits against the largest technology platforms. Because the remedy is structural or behavioral relief—an injunction against the conduct, or in the most serious cases a breakup—rather than money, a conduct case can reshape an entire industry. Section 1 civil cases against horizontal or vertical restraints (where the conduct is not charged criminally) also fall in this family.

Criminal prosecutions are the third and most distinctive family, and they are reserved for the conduct the antitrust laws treat as flatly illegal—hard-core cartels. These are agreements among competitors to fix prices, to rig bids (deciding in advance who will win which contract and at what price), or to allocate markets(carving up customers, territories, or product lines so the conspirators do not compete). Such agreements are per se violations of Sherman Act Section 1: there is no efficiency defense and no inquiry into whether the prices were reasonable, because the conduct is conclusively presumed to harm competition. The Division prosecutes these as crimes, seeking corporate criminal fines and—crucially—prison terms for the individual executives who orchestrated them. This is the enforcement that gives antitrust its teeth, and it is the part of the Division's work that no civil enforcer can replicate.

The criminal authority and the leniency program

The single most important thing that distinguishes the DOJ Antitrust Division from every other competition enforcer in the federal government—and from the FTC in particular—is that it holds criminal antitrust authority. The Sherman Act makes price-fixing, bid-rigging, and market allocation felonies, and only the Department of Justice can prosecute them. The FTC, the other civil antitrust enforcer, has no criminal power; it cannot indict anyone. When a cartel is uncovered, the matter is a DOJ matter by necessity. The consequences for individuals are severe: an executive convicted of a hard-core antitrust conspiracy can be sentenced to prison, and the corporate fines for the largest cartels reach into the hundreds of millions of dollars. This is why the criminal matters in the dataset name not only corporate defendants but individual ones, and why the relief_sought on a criminal matter includes imprisonment.

The engine that makes the cartel program work is the Corporate Leniency Program. Cartels are by their nature secret, and proving a conspiracy that exists only in private meetings and coded communications is extraordinarily hard. The leniency program solves this with a structured incentive: the first conspirator to come forward, confess the conspiracy, and cooperate fully with the investigation can receive complete amnesty—avoiding criminal conviction and fines for the company and prosecution for its cooperating employees—while the conspirators who hesitated face the full force of prosecution. Because only the first applicant wins amnesty, the program creates a race to the prosecutor's door: each cartel member knows that if any other member defects first, it is too late. This destabilizing dynamic has made the leniency program the most productive tool in cartel enforcement worldwide, and it is the reason a substantial share of the Division's criminal matters originate not from independent investigation but from a self-reporting conspirator. The program leaves a signature in the data: clusters of related prosecutions arising from a single industry-wide conspiracy, often with one conspicuously absent defendant—the amnesty applicant.

The division of labor with the FTC

The most common point of confusion about federal antitrust enforcement is that there are two federal enforcers, not one. The DOJ Antitrust Division shares civil antitrust authority with the Federal Trade Commission. Both can challenge mergers under the Clayton Act, and both can bring civil cases against anticompetitive conduct—the FTC primarily under Section 5 of the FTC Act, which prohibits unfair methods of competition, and which reaches conduct that also violates the Sherman and Clayton Acts. The two agencies have overlapping jurisdiction, and to avoid duplicating each other's work they allocate matters by industry expertise through a long-standing clearance process: when a merger is announced, the agencies decide between them which one will review it, generally assigning it to the agency with the greater accumulated expertise in that sector.

The practical result is a rough industry specialization. The Antitrust Division has historically taken the lead in sectors such as telecommunications, financial services, airlines, agriculture, defense, and, increasingly, the largest technology platforms, while the FTC has concentrated in areas including health care, pharmaceuticals, retail, energy, and consumer products—though the allocation is fluid and decided matter by matter rather than by any fixed boundary. The key asymmetry remains: the FTC and DOJ share civil authority, but only DOJ can prosecute criminally. For an analyst, this means the doj_atr record is only one half of the federal civil antitrust story. A complete picture of government merger and conduct enforcement requires joining it to the FTC enforcement record, because a deal challenged by the FTC will not appear in the DOJ data at all—and conversely, a cartel prosecution will never appear in the FTC's. The two datasets are complementary halves of a single enforcement universe, not duplicates.

Hart-Scott-Rodino premerger review

Upstream of every merger matter in the dataset sits a review process that is invisible in the case record but indispensable to understanding it: the Hart-Scott-Rodino Antitrust Improvements Act of 1976, universally called HSR. Before HSR, the antitrust agencies often learned of a merger only after it had closed, by which point unscrambling the combined company was difficult or impossible. HSR fixed this by requiring parties to large transactions to notify the agencies before consummating the deal and to observe a mandatory waiting period—giving the DOJ and the FTC the chance to review the transaction while it can still be stopped cleanly.

The HSR process is a funnel. Transactions above the statutory size thresholds must file a premerger notification with both agencies and wait. The reviewing agency conducts an initial review during the waiting period; the great majority of filings raise no competitive concern and clear without further action. For the small fraction that warrant closer scrutiny, the agency issues a Second Request—a demand for extensive documents and data—which extends the waiting period and triggers an intensive investigation. Only at the end of that funnel, for the handful of deals the agency concludes are anticompetitive and the parties will not fix through divestitures, does a matter become a litigated merger challenge and enter the case record. This is the critical context for the doj_atr data: the cases in it are the visible tip of a much larger review apparatus. Thousands of mergers are notified each year, a small number draw Second Requests, and only a fraction of those become filed cases. The litigation record undercounts the Division's influence enormously, because most of its merger work happens in the HSR review—deals restructured or abandoned under the threat of a suit that was never filed—and leaves no entry in a record of cases.

The landmark cases: from Microsoft to big tech

The case record is also a history of American capitalism told through its largest competition fights. The Division's lineage runs back to the trust-busting era—the breakups of Standard Oil and American Tobacco in 1911 under the Sherman Act, and later the 1982 consent decree that broke up the AT&T telephone monopoly into the regional “Baby Bells”—and forward through the modern era. The matters that define the dataset are the ones that reshaped industries.

United States v. Microsoft, filed in the late 1990s, is the landmark Section 2 monopolization case of the computer age. The government alleged that Microsoft had unlawfully maintained its monopoly in personal-computer operating systems through exclusionary conduct—most famously, tying its Internet Explorer browser to Windows to suppress a rival browser that threatened the operating-system monopoly. The case produced a finding of liability and a remedial decree, and it stands as the template for how the government confronts a dominant technology platform: not by attacking its success, but by challenging the conduct it uses to entrench itself. Its echo is unmistakable in the modern wave of big-tech monopolization suits. In recent years the Division has brought major Section 2 cases against Google—one concerning the firm's dominance in general search and the distribution agreements that entrench it, and another concerning its control of the advertising-technology stack that sits between publishers and advertisers—the most consequential monopolization litigation since Microsoft.

Alongside the conduct cases run the major merger challenges—suits to block large combinations in airlines, telecommunications, agriculture, publishing, and other concentrated industries—and the steady output of the criminal cartel program, which over the decades has prosecuted price-fixing and bid-rigging conspiracies across industries from auto parts to financial benchmarks to commodity chemicals, generating some of the largest corporate criminal fines in any area of federal law. Read together, these matters trace the evolution of what the government considers a threat to competition: from the physical trusts of the industrial age, to the operating-system gatekeeper of the desktop era, to the platform intermediaries of the digital economy.

Joining to the FTC and the broader DOJ record

The doj_atr table is most valuable when it is read as one facet of the federal enforcement record rather than in isolation, and the defendant names are the join key that connects it to the rest. Three linkages matter most.

The first is the join to the FTC enforcement record, the other federal antitrust enforcer. Because civil antitrust authority is shared, a complete accounting of government competition enforcement requires both datasets: a merger the FTC reviewed and challenged lives in the FTC record, a merger DOJ challenged lives here, and only the union of the two answers a question like “how many large mergers did the federal government challenge in a given period?” Joining on defendant names also surfaces the occasional firm that draws scrutiny from both agencies across different matters, and it lets an analyst compare the two enforcers' merger and conduct postures over time—something neither dataset can support alone.

The second is the join to the broader DOJ prosecution record. The criminal antitrust matters are a specialized slice of the Department's overall criminal docket, and a defendant prosecuted for a bid-rigging conspiracy may also appear in the general prosecution data, often charged alongside related fraud or obstruction offenses that accompany a cartel. Linking the criminal antitrust matters to the wider prosecution record places cartel enforcement in the context of the Department's total white-collar effort and reveals how often an antitrust conspiracy travels with other federal crimes.

The third linkage is to the compliance and screening ecosystem. A criminal antitrust conviction—particularly for bid-rigging on government contracts—is exactly the kind of integrity failure that can lead to suspension or debarment from federal contracting, so a defendant in the doj_atr record may reappear on the federal exclusions list. For any organization that screens its counterparties against the full battery of federal enforcement lists, the antitrust case record is one of the inputs: a firm or executive named in a price-fixing prosecution is a counterparty that a risk-aware compliance program wants to know about.

Analytical uses

A structured, decades-long record of every government antitrust case supports analysis that no single matter, and no narrative account, can provide.

Enforcement-intensity trends over time is the most immediate use. Tallying matters by filing year and by type—merger, civil conduct, criminal—reveals the rhythm of antitrust enforcement across administrations and decades: the periods of aggressive merger challenge and the periods of restraint, the rise and fall of the criminal cartel caseload, and the recent resurgence of Section 2 monopolization litigation after a long quiet stretch. Because antitrust policy swings with the priorities of successive administrations more than almost any other area of federal enforcement, the time series is a direct measure of how the government's appetite for competition enforcement has changed.

Industry and sector concentration of enforcement exploits the industry field. Aggregating matters by sector shows where the government has focused its competition concern—telecommunications and airlines in some eras, technology platforms in others—and how that focus has migrated as the economy itself has changed. Cross-referenced with the DOJ-FTC allocation, it also illuminates the practical boundary between the two agencies' territories. Outcome analysis uses the disposition field to ask how the government's cases actually end: the share resolved by consent decree versus litigated to judgment, the frequency with which sued deals are abandoned, and the conviction and plea record in the criminal matters—a window into the Division's win rate and its reliance on settlement. Finally, defendant and recidivism analysis uses the named-party fields to identify firms and executives that appear more than once, the corporate families that recur across cartel prosecutions, and the cross-enforcer overlap with the FTC record.

Python workflow: classifying and tallying the case record

The script below pulls the Antitrust Division's case announcements from the public justice.gov feed, classifies each matter as a merger challenge, a civil conduct case, or a criminal prosecution by the language of its announcement, tallies the matters by type and by year, and surfaces the matters that name major defendants. No API key is required for the public data. Because the Division's case record is published as a mix of structured case filings and free-text announcements, the classification here is heuristic—keying on the vocabulary of merger, monopolization, and cartel enforcement—and any production use should be validated against the authoritative case-filings listing and refined with the structured statute and case-type fields where they are available.

import requests, re
import pandas as pd
from collections import Counter

# DOJ Antitrust Division case filings and announcements.
#
# The Division publishes its case record on justice.gov/atr -- the
# "Antitrust Case Filings" listing and the press-release feed of case
# announcements. Both are public and require no API key. justice.gov
# also exposes a JSON view of its press-release content type, which is
# the most stable machine-readable surface for the announcements that
# accompany merger challenges, civil-conduct suits, and criminal
# prosecutions.
#
# This script:
#   1. Pages the Antitrust Division press feed into a DataFrame
#   2. Classifies each matter as merger / civil-conduct / criminal
#   3. Tallies matters by type and by year
#   4. Surfaces the largest matters by the named defendants
BASE = "https://www.justice.gov/api/v1/press_releases.json"
ATR_COMPONENT = "antitrust-division"   # component slug on justice.gov


def fetch_atr(pages=40, page_size=50):
    rows = []
    for p in range(pages):
        params = {
            "parameters[component]": ATR_COMPONENT,
            "pagesize": page_size,
            "page": p,
        }
        r = requests.get(BASE, params=params, timeout=60)
        if r.status_code == 404:
            break
        r.raise_for_status()
        batch = r.json().get("results", [])
        if not batch:
            break
        rows.extend(batch)
        print(f"  fetched {len(rows):,} announcements so far...")
    return pd.DataFrame(rows)


# --- Classify a matter by the language of its announcement -----------
MERGER = re.compile(r"\b(merger|acquisition|acquire|divest|Section 7|block the|"
                    r"proposed (deal|transaction))\b", re.I)
CRIMINAL = re.compile(r"\b(price[- ]fix|bid[- ]rig|market allocation|cartel|"
                      r"indict|plead|sentenc|conspiracy|criminal)\b", re.I)
CIVIL_CONDUCT = re.compile(r"\b(monopoliz|Section 2|exclusionary|tying|"
                           r"restraint of trade)\b", re.I)


def classify(text):
    t = text or ""
    if CRIMINAL.search(t):
        return "criminal"
    if MERGER.search(t):
        return "merger"
    if CIVIL_CONDUCT.search(t):
        return "civil_conduct"
    return "other"


df = fetch_atr()
print(f"Antitrust Division announcements loaded: {len(df):,}")

# justice.gov field names vary; resolve title/body/date defensively.
def col(frame, *cands):
    low = {c.lower(): c for c in frame.columns}
    for c in cands:
        if c.lower() in low:
            return low[c.lower()]
    return None

c_title = col(df, "title")
c_body  = col(df, "body", "teaser")
c_date  = col(df, "date", "created", "changed")

df["_text"]  = (df[c_title].fillna("") + " " + df[c_body].fillna("")) if c_body else df[c_title]
df["_type"]  = df["_text"].map(classify)
df["_year"]  = pd.to_datetime(df[c_date], errors="coerce").dt.year

# --- 1. Matters by type ---------------------------------------------
print("\nMatters by type:")
for t, n in df["_type"].value_counts().items():
    print(f"  {t:<16} {n:>5,}")

# --- 2. Matters by year ---------------------------------------------
print("\nMerger vs civil-conduct vs criminal by year:")
pivot = (df[df["_type"] != "other"]
         .pivot_table(index="_year", columns="_type",
                      values=c_title, aggfunc="count", fill_value=0))
print(pivot.tail(10).to_string())

# --- 3. Largest matters by named defendants -------------------------
# A crude proxy for "largest": announcements naming well-known firms.
WATCH = ["Google", "Microsoft", "Apple", "Visa", "Live Nation",
         "JetBlue", "American Airlines", "UnitedHealth"]
print("\nMatters naming major defendants:")
for firm in WATCH:
    hits = df[df["_text"].str.contains(re.escape(firm), case=False, na=False)]
    if not hits.empty:
        first = pd.to_datetime(hits[c_date], errors="coerce").min()
        print(f"  {firm:<18} {len(hits):>3} announcement(s); "
              f"earliest {first.date() if pd.notna(first) else 'n/a'}")

Two practical notes apply. First, the classification is deliberately language-based and therefore approximate: an announcement that mentions both a merger and a related criminal charge will be assigned by the precedence rules in the classify function, and a terse filing notice may fall into the “other” bucket. A rigorous analysis should anchor the type on the structured case_type and statute fields from the case-filings record rather than inferring it from prose, using the text classification only to fill gaps. Second, for serious longitudinal work—tracing enforcement intensity across administrations, or building the cross-enforcer merger picture—the better foundation is the Division's structured case-filings listing joined to the FTC competition record, which together cover the whole of federal civil antitrust enforcement rather than the announcements of a single agency.

Limitations and analytical caveats

The antitrust case record is the authoritative public account of government competition litigation, but it is a record of cases, and that framing carries structural limits an analyst must internalize before drawing conclusions.

The litigation record undercounts enforcement. The most important caveat follows directly from the HSR process: the cases in the data are the small fraction of matters that actually became filed suits. The vast majority of the Division's merger work happens before any case is filed—deals cleared after review, deals abandoned under the threat of a challenge, deals restructured through divestitures negotiated during the HSR waiting period. None of that leaves a case entry. A count of filed merger challenges therefore measures only the visible tip of the Division's influence and will badly understate how many transactions its review actually shaped. The same is true of conduct matters that close after investigation without a suit. The dataset records what was litigated, not what was enforced.

Enforcement intensity is a policy choice, not a measure of competition. Antitrust enforcement swings sharply with the philosophy of each administration—periods of vigorous challenge alternate with periods of deliberate restraint. A low count of cases in a given span may reflect a permissive enforcement posture rather than a healthy, competitive economy, and a surge may reflect a more interventionist policy rather than a sudden outbreak of anticompetitive conduct. The case count measures the government's decisions to act, which are driven as much by changing legal and economic theory as by underlying market conditions. Treating the volume of cases as a thermometer of market competition reverses the causation the data can actually support.

A challenge is not a finding, and an outcome is not always clear. A filed case alleges a violation; it does not establish one. Merger challenges frequently end in consent decrees or abandoned deals rather than judicial findings of illegality, and the parties to a consent decree typically admit no wrongdoing. Conduct cases can be litigated for years and produce mixed or partial outcomes. Reading the existence of a matter as proof that the defendant violated the antitrust laws over-reads the record. The outcome and disposition_date fields capture the resolution imperfectly: a deal that was abandoned the day after the suit was filed and a case litigated to a full liability judgment are very different events that a coarse outcome code may not distinguish.

It is half the federal picture, and entity resolution is hard.Because civil antitrust authority is shared with the FTC, the doj_atr record on its own is an incomplete account of federal competition enforcement—every FTC merger and conduct case is missing—and any analysis of “federal antitrust enforcement” that uses only this dataset will systematically omit whole sectors that the FTC handles. Finally, the matters are keyed by defendant names rendered in free text, and the same firm can appear under slightly different corporate names, after a rename, or as a parent versus a subsidiary; joining the case record to itself for recidivism analysis, or to the FTC and broader prosecution records, depends on careful entity resolution rather than naive string matching.

Held with these caveats in mind, the doj_atr table is a uniquely valuable record: roughly 920 matters in which the United States went to court—to block a merger, to restructure a monopolist, or to imprison a price-fixer—tracing more than a century of how the federal government has tried to keep American markets competitive, from the trusts of the industrial age to the platforms of the digital one.