The inspector does not begin by asking what the factory would pay. He asks what has come out of the pipe. A limit has been written, a sample is taken, and the result is compared with the rule. In another factory he may ask whether the required equipment has been installed, or whether a prohibited substance has been used at all. These are all called standards, though they leave very different decisions to the person regulated. Their common promise is that some conduct will be kept within a boundary. The economist's question is what that promise costs, and whether a price could secure the relevant protection more cheaply. The public's question is what protection would remain if the factory could simply pay.
Begin by distinguishing the things a rule can fix. A technology requirement names a means: install this apparatus, use this process. A performance limit names a result and may leave the means open. An emissions ceiling states how much may leave a source; an ambient standard concerns what may be present where people or other things are exposed. Even a performance rule can be expressed as a rate rather than a total, allowing aggregate emissions to rise as production expands. The machinery of inspection, the freedom to innovate and the certainty actually offered all depend on which of these has been specified. The phrase command and control conceals too much if it makes every rule resemble an order to buy the same filter.
The rate and the total can diverge without anyone violating the rate. In an invented factory, two units of discharge per unit of output, at one hundred units of output, produce two hundred units of discharge. A rule halves the permitted rate, but output subsequently triples: the factory now discharges three hundred. Its product has become cleaner while its total contribution has grown. A total ceiling of one hundred would ask something different; an ambient constraint would ask how the discharge affects the receiving place. Before comparing a rule with a price, one must identify the outcome to be protected. Certainty about the wrong number is no protection against the harm that number omits.
A uniform obligation can be expensive because plants differ. The earlier Permit makes the point by shifting a fixed amount of abatement towards the plant able to provide it more cheaply. A charge on each residual emission gives every plant a reason to find reductions costing less than the charge; tradable allowances can perform the same sorting beneath an aggregate cap. A regulator could also assign different individual limits that reproduce the least-cost allocation if it knew the relevant costs and could keep its instructions abreast of them. The advantage of exchange is that the necessary adjustment can arise from the plants' own choices. The argument concerns the allocation of a specified burden; it does not establish that every individual limit is wasteful.
Nor need the firm be told how to comply. A ceiling on the measured discharge can leave it free to change fuel, redesign production, recover a waste product or install treatment. Naming a particular machine can close some of those routes, though it may simplify an inspection where the result is difficult to observe. The choice is then between kinds of knowledge. The authority may know that a certain device works under specified operating conditions without being able to measure every release continuously; the operator may know a cheaper way whose performance the authority cannot yet verify. A good rule must decide what evidence will demonstrate equivalent protection. Otherwise yesterday's approved machine can become a barrier to tomorrow's improvement, while a promise of innovation can become an excuse for performance that nobody can check.
The cost of knowing and enforcing belongs inside the comparison. A unit charge requires a credible quantity on which to levy it. A prohibition may sometimes require only a reliable test that a substance is present; a performance ceiling may need almost the same measurement as a tax. Conversely, observing a filter does not establish that it is being operated, maintained or used in the conditions that make it effective. Weitzman noted the possibility that a pass-or-fail test could be cheaper than measuring output on a continuous scale. That observation supplies a reason to compare particular instruments, not a presumption that every prohibition is easy to enforce.
A rule is certain about what it requires. Protection depends on what it measures and whether the requirement is met.
Written severity and actual compliance can come apart. Suppose avoiding compliance saves a risk-neutral operator one hundred, the only consequence of detection is a two-hundred fine, and the chance of detection is one fifth. The expected fine is forty. Under these invented assumptions, disobeying pays. Raise the probability to three fifths and the expected fine is one hundred and twenty; the calculation reverses. This is not an estimate of any regulator's effectiveness, and real sanctions may include closure, remedial orders, reputation or personal liability. It is a reminder that printing a maximum discharge does not physically constrain a pipe. A tax that can be evaded and a permit that can be counterfeited face the same distance between the rulebook and the world.
A local limit may also protect something a broad price deliberately leaves flexible. If an extra discharge at one outfall produces a dangerous local concentration, allowing that source to pay the same price as a remote source can miss the relevant constraint. The Permit showed why equal tons need not mean equal exposures. One response is a source-specific performance limit; another is a differentiated charge or a trading rule whose exchanges preserve the local requirement. The choice turns partly on which relationship can be measured and administered. The case for a limit is strongest when the protected boundary is important and a rule can reliably enforce it; the adjective strict alone supplies neither condition.
A price and a standard can therefore work together. Baumol and Oates proposed beginning with a chosen environmental-quality requirement and using charges adjusted to attain it. The resulting target was not certified as the welfare optimum merely because it could be reached economically. At a different level, an aggregate trading scheme can leave individual health-protection limits in place. EPA's 1993 explanation of sulfur-dioxide allowance trading explicitly said that holdings did not permit a unit to violate federal atmospheric standards protecting public health. Such arrangements divide the work: a boundary protects a specified interest, and prices guide choices within the feasible space. Their merits must be judged from the combined obligations, rather than by giving the whole arrangement the name of its most conspicuous instrument.
The instruments also give different reasons to improve beyond compliance. Under a bare individual ceiling, an extra reduction below the limit may earn no regulatory saving unless credits, future obligations or some other benefit attach to it. A charge continuing on residual emissions rewards every further measured reduction at the rate charged; a permit market can reward a reduction by releasing an allowance for sale. A technology rule may still induce an inventor to build a cheaper approved device, and a performance rule may induce a cheaper way to reach its limit. The narrower point is that the reward for being cleaner than required must be found somewhere. It cannot be assumed to exist merely because the original requirement was demanding.
The choice posed by the plate is answered at the boundary of a particular harm. If the aim is to reach a shared target cheaply, unequal abatement costs make room for a common price or for trade. If the aim includes a local limit, a prohibited exposure or a result that can only be reliably tested in a certain way, the instrument must preserve that protection. Both choices require measurement, compliance and a judgment about who bears the cost. The standard states what the community requires; the price offers a way to economise while pursuing it, or an alternative way of directing conduct. Neither can avoid the question left for the volume's last plate: which consequences belong within the account we have decided to regulate?