The Problem with Every New Law
Every law and regulation is justified by a problem it claims to solve. But watch what happens after the law passes. The problem disappears from the conversation. The goal is forgotten. What remains is the permanent bureaucracy, the permanent funding, the permanent power. The machine that was supposed to fix a specific issue becomes a permanent feature of the landscape, regardless of whether the issue was ever fixed.
This is no accident. It is the natural result of a legislative process that never asks the questions that matter. The people who propose laws have no incentive to define success, because success would mean the law could be retired. The people who implement laws have no incentive to measure results, because bad results would threaten their budgets. The people who benefit from laws have no incentive to include sunset clauses, because sunsets mean the money stops.
The solution is not to propose different policies. The solution is to change how policies are designed. Every law should answer seven questions before it is enacted. If it cannot answer them, it should not pass.
Question 1: What Problem Are You Solving?
This is the only question that gets asked, and even it is usually answered poorly.
The problem must be stated in concrete, measurable terms. Not “we need to address climate change” but “residential electricity emissions were X million tonnes in the baseline year and we want them to be Y million tonnes within Z years.” Not “we need to make housing more affordable” but “the median rent-to-income ratio in metropolitan areas is currently 32 percent and we want it to be 25 percent within ten years.”
A vague problem statement is a warning sign. It means the goalposts can be moved later. When the policy fails to deliver, the proponents can say “we never said we would reduce rents - we said we would address affordability” - a word that means whatever they need it to mean at any given moment.
The problem must also be quantified in a way that can be independently verified. Government departments should not be the sole source of the data used to measure their own performance. The baseline must be established by an independent body, or at minimum by a methodology that the public can replicate.
Question 2: Who Wins and Who Loses?
Every policy creates winners and losers. There are no exceptions. A policy that claims to help everyone is either lying or so vague that it helps no one.
The winners are usually obvious, because they are the ones advocating for the policy. The losers are usually invisible, because they are diffuse, unorganized, and often unaware that the policy will harm them until after the harm occurs. A tariff helps domestic steel producers - a small, organized group that can lobby effectively. It hurts every consumer who buys anything made with steel - a huge, diffuse group that will never organize a protest over a few pennies on a can of beans.
The obligation of honest policymaking is to name the losers explicitly before the law passes. Not in a footnote in a regulatory impact assessment that nobody reads. In the legislative debate, on the record, in plain language. “This law will increase the cost of housing construction by 7 percent. Renters will pay for this in higher rents. Here is why we think the benefit is worth that cost.”
If the proponents cannot name who loses, they have not thought hard enough about what the policy actually does.
Question 3: How Will This Be Implemented?
This is where most policies die if anyone bothers to ask.
Implementation is not “we will create a new agency to oversee X” or “we will task the existing department with developing guidelines.” Implementation is the specific mechanism by which a change in the real world occurs. Who will do what, with what authority, using what resources, on what timeline?
A carbon tax is easy to implement because it uses an existing mechanism - the tax system. You change a number in a spreadsheet and the price signal propagates automatically. A cap-and-trade system is hard to implement because it requires creating a new market, a new regulator, new monitoring systems, and new compliance procedures. A zoning reform is hard to implement because it requires changing the behavior of thousands of local governments, each with its own political dynamics and its own ability to delay.
Every additional step between the law and the outcome is an opportunity for the outcome to be diluted. Implementation must be designed with the knowledge that the people implementing it may not want it to succeed. Bureaucrats who lose power under a new system will find ways to slow it down. Industries that are harmed will find ways to exploit loopholes. Courts that are asked to interpret vague language will interpret it in ways the drafters did not intend.
The question is not “can this law be passed?” The question is “can this law be implemented as designed, against the resistance of everyone who loses from it?”
Question 4: How Will You Measure the Effects?
The measurement must be independent, pre-registered, and public.
Before the law takes effect, the metrics must be established. The baseline must be recorded. The methodology for calculating the metrics must be specified in enough detail that an independent researcher could replicate it. The timeline for measurement must be set - one year, five years, ten years - with intermediate checkpoints.
This sounds obvious. It is almost never done.
The UK’s 2010 Health and Social Care Act was one of the most significant pieces of domestic legislation in a generation. It reorganized the entire National Health Service. The government’s own impact assessment could not quantify the expected benefits. The stated goals - greater efficiency, better patient outcomes, more accountability - were never defined in measurable terms. Ten years later, no one could say whether the reforms had worked, because the question had never been operationalized.
This is not a failure of implementation. It is a feature of the design. If you cannot measure the effect, you cannot prove the policy failed.
Question 5: How Will You Report the Results to the Electorate?
Reporting must be mandatory, periodic, and published in plain language.
Every year on the anniversary of the law’s enactment, the responsible department must publish a report that states:
- The original problem and the target
- The current state of the metrics
- Whether the policy is on track to meet its goals
- Any adjustments made during the year
- The total cost to date, including implementation and compliance costs
The report must be written for a general audience. No acronyms, no jargon, no charts that require a graduate degree to interpret. It must be published on a single webpage that is linked from every government statement about the policy. It must be submitted to the legislature for debate.
This is the closest equivalent to a publicly traded company’s quarterly earnings report. If a company can report its financial performance every three months with enough detail for investors to make decisions, a government department can report a policy’s performance once a year with enough detail for voters to decide whether it is working.
Question 6: What Does Success Look Like?
Success must be defined before the law takes effect, in the same measurable terms as the problem.
If the goal was to reduce residential electricity emissions by 20 percent within ten years, success is a 20 percent reduction within ten years. If emissions fall by 10 percent, the policy partially succeeded. If they rise, it failed. If they fall by 20 percent but the cost was five times the estimate, the policy may have succeeded on its own terms but failed as a matter of public policy.
Success is not “we have started the conversation” or “we have raised awareness” or “we have laid the groundwork for future action.” These are phrases that people use when they cannot point to a concrete result. Success is a number, a date, and a cost. If you cannot state all three before the policy begins, you do not know what you are trying to achieve.
Question 7: What Is the Sunset Clause?
This is the most important question and the one that is almost never asked.
A sunset clause specifies when the law will expire unless it is explicitly renewed. It forces a periodic reassessment. If the problem is solved, the law dies naturally. If the policy is not working, the law dies and a new approach can be tried. If the policy is working, renewal requires a debate in which the evidence is presented and the opposition has a chance to argue that the law should be allowed to expire.
Sunset clauses are common in some areas of law - the USA PATRIOT Act had them, the EU’s sanctions regimes have them, many temporary tax provisions have them. They are almost never used for the kind of permanent regulatory expansion that characterizes modern government.
The objection to sunset clauses is that they create uncertainty. Businesses cannot plan if they do not know whether a regulation will exist in five years. This objection is valid but weak. The alternative - permanent regulation that never dies, never gets reviewed, and never gets retired - creates its own uncertainty: the certainty that bad policy will persist forever.
A ten-year sunset is standard for most regulations. Five years for experimental or contested policies. Twenty years for major infrastructure or institutional reforms. The length should be proportional to the investment required to comply, but it should always exist.
Applying the Questions
Consider any major piece of legislation from the last twenty years. Apply the seven questions. See how many have acceptable answers.
The carbon tax debates across Europe: the goal was clear (reduce emissions), but the winners and losers were never honestly named (drivers lost, the treasury won). The implementation was through the existing tax system, which was efficient, but on top of existing fuel duties that already priced carbon at twice the social cost - so the “problem” it was solving had already been solved by a tax nobody remembered was there. The measurement was outsourced to climate models that could not isolate the effect of the tax from other factors. The reporting was buried in budget documents. Success was never defined in terms of the tax’s specific contribution. Sunset clauses were never proposed.
The UK’s apprenticeship levy: the goal was to increase the number of high-quality apprenticeships. The winners were training providers and large employers who could absorb the cost. The losers were small businesses that could not. Implementation was through a payroll tax with a credit system so complex that thousands of employers simply wrote off the money. Measurement showed that the number of apprenticeships increased but their quality decreased - a result that the levy’s design should have predicted. Success was never defined, because the government’s target was “three million apprenticeships by 2020” - a quantity target that incentivised short, low-quality placements. The levy has no sunset clause. It will exist forever, regardless of whether it works.
The US Inflation Reduction Act: the goal was to accelerate clean energy deployment. The winners were solar and wind manufacturers, electric vehicle producers, and battery companies. The losers were taxpayers (through higher debt and inflation), coal and gas communities, and anyone who would have preferred a technology-neutral approach. Implementation was through a combination of tax credits, grants, and loan guarantees - a layer cake of complexity that has already produced fraud, controversy, and legal challenges. Measurement is entangled with other climate policies, making the IRA’s specific contribution impossible to isolate. Success was defined as “net emissions 40 percent below 2005 levels by 2030,” but the IRA is only one of many policies contributing to that target, so even if the target is met, the IRA’s specific value cannot be known. No sunset clause.
Every one of these policies was enacted with good intentions. Every one fails at least five of the seven questions.
Why They Will Not Adopt This
The seven-question framework is not technically difficult. It does not require new technology, new agencies, or new funding. It requires a piece of paper and the willingness to be honest.
That is why it will not be adopted.
The current system exists precisely because it does not answer these questions. Vagueness allows politicians to claim credit for popular goals without being held accountable for results. The absence of measurements allows departments to continue programs that are not working. The lack of sunset clauses allows the regulatory state to expand permanently.
Every person who benefits from an existing policy will oppose a framework that subjects their policy to periodic review. The majority of political energy in any government is spent not on creating new policies but on defending existing ones from scrutiny. The seven questions would threaten that defence.
This does not mean the framework is naive. It means the framework would work exactly as designed - and the people who benefit from the current system would fight it.
The Question
This is not a radical proposal. It is the minimum standard for accountable government. A publicly traded company faces more rigorous reporting requirements than a government department running a billion-pound program. A pharmaceutical company must demonstrate that its product works through randomised controlled trials before it can be sold. A government can impose a regulation on millions of people based on a white paper and a press conference.
The seven questions would not prevent any policy that can honestly answer them. They would only prevent policies whose proponents cannot bring themselves to say what they are trying to achieve, who will lose, how it will work, how they will know if it worked, what success looks like, and when the experiment will end.
If a policy cannot survive those questions, it should not survive at all.
Are you willing to ask these questions of every law that claims to help you? Or would you rather not know the answers?