A woman applies for a disability benefit and is refused in a two-line letter. A restaurant is shut by a health inspector the day before a holiday. A mining company is told its permit is cancelled. A journalist asks a ministry for the contract behind a new bridge and gets silence. None of these decisions came from a parliament or a judge. Each was made by an official acting under a law, and each can change a life.
Modern governments do thousands of such things every day, because statutes are written in broad terms and someone must apply them to the individual case. So the question for lawyers is not whether officials may decide, but how they can be made to stay inside the law: what is the power, what is the procedure, who can challenge it, and who can look at the papers? That is administrative law. A graduate keeps seeing a pattern in every news story about an agency: power, limits, review.
This describes how legal systems work and isn't legal advice. Laws, deadlines and courts differ by country and change. If a government decision affects you, check the time limits for appeal with a lawyer or advice service where you live.
The human problem is as old as rulers: someone with power makes decisions about people who cannot easily answer back. The common solution is the rule of law: officials may do only what a law lets them do, and a person affected can go to an independent body. A modern state also needs to keep its agencies expert and quick. All systems balance those two aims.
Where they differ is who does the reviewing. Germany's Basic Law is blunt: "Should any person's rights be violated by public authority, he may have recourse to the courts" (Article 19(4)). Brazil's constitution is even more sweeping: the law may not exclude from review by the judiciary any injury or threat to a right (Article 5, XXXV). The question is which courts.
| Family | Who reviews officials | Distinct features |
|---|---|---|
| United States | Ordinary federal courts, under the Administrative Procedure Act of 1946 and the Constitution. | Courts set aside action that is "arbitrary and capricious" or beyond statutory power. Judicial review of government itself dates to Marbury v. Madison (1803). |
| England and Wales | The Administrative Court and tribunals, by "judicial review". | Largely developed through case law; reasonableness, fairness and legality; the Human Rights Act 1998 adds a rights-based layer. |
| France and continental Europe | Separate administrative courts, topped in France by the Conseil d'État; Germany has its own administrative court branch. | A body of public-law rules distinct from private law; judges trained in public law. |
| Latin America | Mixed. Brazil and Argentina use the ordinary judiciary; Colombia has a Consejo de Estado at the top of a separate administrative jurisdiction; Mexico has a federal administrative tribunal alongside the amparo. | The amparo (and Colombia's tutela) is a fast procedure to protect constitutional rights against officials and sometimes private actors. |
Seen from a distance the divide is between systems with one set of courts for everything and systems with a second court branch specialised in the state. Convergence is the story: European Union law, the European Convention on Human Rights and inter-American human-rights law have pushed every system toward similar standards.
When a court looks at a decision it does not usually ask "was this the best decision?", which would make the judge the official. It asks whether the decision was lawful. English law, in the House of Lords' GCHQ case (1985), grouped the grounds into illegality (acting beyond the power the law gives), irrationality (a decision so unreasonable that no reasonable authority could have made it, the old Wednesbury test of 1948) and procedural impropriety (unfair process). US law uses similar ideas: a decision can be "arbitrary and capricious" if the agency ignored an important factor or offered no reasoned explanation (Motor Vehicle Manufacturers v. State Farm, 1983). A fourth ground has spread across the world: proportionality, below.
Procedure is its own ground. A person facing a decision should know the case against them, be able to answer it, and get reasons. The US Supreme Court's test in Mathews v. Eldridge (1976) weighs three things: the private interest, the risk that the procedure gets it wrong (and what extra safeguards would add), and the government's burden. The tool below checks your instincts.
Invented examples. Real cases often raise several grounds at once, and the labels vary by country.
The hottest argument in the field is how much a court should defer. Agencies have expertise, accountability to elected leaders, and the day-to-day problems in front of them. Courts have independence and the job of saying what the law means. In the US the Chevron decision (1984) told courts that where a statute is ambiguous, they must accept the agency's reasonable reading. On 28 June 2024, in Loper Bright Enterprises v. Raimondo, the Supreme Court overruled Chevron (6 to 2; Justice Jackson took no part): courts must use their own judgment about what the statute means, giving an agency's view only the weaker respect that its persuasiveness earns (the Skidmore approach). Earlier decisions made under Chevron were not overturned.
Supporters of Loper Bright say it returns the meaning of the law to judges and stops agencies enlarging their own power. Critics say it hands policy to judges and makes regulation slower and less expert. Other countries have their own mix: Many other courts leave expert and policy judgments to the agency while checking the legal framework.
A toy: readings sit on a 0 to 10 line; the agency's reading is upheld if it lies within a window around the court's own (0 with no deference, 1 with some weight, 3 with strong deference). Real law has no such scale. It illustrates the mechanism only.
When a measure limits a right, many legal systems ask four questions in order. Is the aim legitimate? Is the measure suitable, does it actually help? Is it necessary, is there a gentler way that works nearly as well? Is it proportionate in the narrow sense, do the benefits outweigh the harm to the right? The idea grew in German administrative courts in the late 19th century (the Prussian Higher Administrative Court limited police discretion this way) and was developed by the Federal Constitutional Court from the 1950s, famously in its 1958 Pharmacy decision. Canada adopted a version in R v. Oakes (1986), the Court of Justice of the EU and the European Court of Human Rights use it routinely, and constitutional courts in many countries, including Colombia and Brazil, use proportionality-style tests.
The United States is the notable outlier: its constitutional doctrine mostly uses "tiers of scrutiny" (rational basis, intermediate, strict) rather than proportionality, though proportionality appears in Eighth Amendment cases on punishments. Defenders of proportionality say it forces reasoning into the open. Critics, including some of its own scholars, say the last step asks judges to weigh things that cannot be weighed, and that it can invite judges to substitute their values for the legislature's.
A toy to show the order of the steps; courts do not use numbers, and the aim is assumed legitimate. Suitable if the measure scores at least 3; fails the necessity step if a gentler measure does at least 80% as well; balanced if benefit (aim × effectiveness ÷ 10) is at least the harm.
With the defaults the measure passes the first two steps (the aim is legitimate and the measure helps) and passes necessity (a gentler option does only half as well). It fails at the last step: benefit is 7 × 6 ÷ 10 = 4.2, against a harm of 5. Slide the numbers and watch which step stops it first. The order matters: a measure that fails an early step never reaches the balancing, which is the step most open to argument.
France built a different answer in the aftermath of the Revolution. Revolutionary lawmakers distrusted ordinary judges, so they forbade them from interfering with the administration. The state's own Conseil d'État, created by Napoleon's government in 1799 to advise on laws and settle disputes, grew into a real court. A law of 1872 gave it the power to decide cases itself; its case law shaped modern public law, including state liability (the Blanco decision, 1873) and the tool of the recours pour excès de pouvoir, an open, cheap action to annul any administrative act that breaks the law. Today it is both the government's adviser on draft laws and the top administrative court, with lower administrative courts since 1953 and administrative appeal courts since 1987.
Does the arrangement work? Supporters say judges steeped in public administration are expert and firm, and that the action to annul is open to anyone. Critics worry about a court that is also an adviser to the state. Germany has its own branch of administrative courts, separate from its ordinary courts; the English and American systems chose not to separate. The lesson is not that one design is correct, but that every country must decide how to make the state judge itself less.
A simplified picture of the usual route for challenging an administrative decision. Real routes depend on the subject (tax, immigration, benefits), have time limits, and are changed by reform.
Courts are slow and expensive, so most systems add cheaper watchdogs. The ombudsman investigates complaints against officials and recommends fixes. Sweden created the first modern one in 1809; Finland followed in 1919, Denmark in 1955, New Zealand in 1962 and the United Kingdom's Parliamentary Commissioner in 1967. Spain's Defensor del Pueblo came with its 1978 constitution, Mexico's national human-rights commission (CNDH) in 1990, and the European Union has its own Ombudsman, established in 1995. Most can recommend but not command, so their power rests on reputation and on officials' wish not to be named.
The other tool is the right to see the papers. Sweden's 1766 law on freedom of the press is the oldest, giving public access to official documents. Finland followed in 1951, the US passed its Freedom of Information Act in 1966, and France its law on access to administrative documents in 1978. The world wave came later. Mexico passed a federal law in 2002; India's Right to Information Act and Germany's federal law both date from 2005; Chile's access law took effect in 2009; Brazil's detailed access law is from 2011 and Argentina's national law from 2016. A dozen countries had such laws in 1990 and more than 85 by 2011, according to one count. In 2006 the Inter-American Court of Human Rights recognised access to state-held information as part of freedom of expression, in Claude Reyes v. Chile.
Such laws can also be undone. In 2025 Mexico dissolved its autonomous transparency institute (INAI) and moved its functions into a new anticorruption secretariat of the executive branch. The government said transparency would be preserved, and civil-society groups warned that the body deciding whether to release information was now part of the government itself. It is a live argument, not a settled one.
Twelve countries picked for this piece, with the year of a first general national access law as given in the sources (Brazil: the 2011 procedures law; the UK: in force 2005; Chile: in force 2009). Not a world count, and "first law" is not the same as an effective one.
Government acts through officials, so the law of government is the law of officials: the power must come from a law, the process must be fair, the reasons must make sense, and a person must be able to challenge it. Countries differ on which courts review, how much they defer, and whether they ask the proportionality questions. Watchdogs and open-records laws add cheaper checks, and all of these can be strengthened or weakened by politics.
Try this: next time an agency does something in the news, name the power it claims, the process it followed, who could challenge it and where, and whether the public can see the papers.
Sources. Loper Bright Enterprises v. Raimondo (28 June 2024; 6-2; overruled Chevron (1984)); Basic Law Art. 19(4) (English translation, Federal Ministry of Justice); Brazil Constitution Art. 5 XXXV; Conseil d'État history (1799; 1872; tribunaux administratifs 1953; cours administratives d'appel 1987), per Wikipedia and public sources; Blanco (1873), Wednesbury (1948), GCHQ (1985), Mathews v. Eldridge (1976), State Farm (1983), Marbury (1803), the Pharmacy decision (1958) and R v. Oakes (1986) are standard case-law landmarks taught in every public-law syllabus. Origins of proportionality in Prussian administrative courts: summary on Wikipedia, "Proportionality (law)". Ombudsman dates: Wikipedia, "Ombudsman". Freedom-of-information dates: Wikipedia, "Freedom of information laws by country" (Sweden 1766, Finland 1951, France 1978, Colombia 1985, Brazil 1991 and 2011, Chile 2009, Argentina 2016, India and Germany 2005); US FOIA 1966, UK FOIA 2000, Mexico 2002 are standard references. Mexico's INAI: Mexican press reports of March 2025 (Expansión, Por Esto, Excélsior). The 2006 Inter-American Court judgment: Claude Reyes v. Chile.