A simple theory of regulations, new and old

Q = min (#laws, #regulators)

The number of laws grows rapidly, yet the number of regulators grows relatively slowly.  There are always more laws than there are regulators to enforce them, and thus the number of regulators is the binding constraint.

The regulators face pressure to enforce the most recently issued directives, if only to avoid being fired or to limit bad publicity.  On any given day, it is what they are told to do.  Issuing new regulations therefore displaces the enforcement of old ones.

If the best or most fundamental regulations are the ones issued first, over time the average quality of regulation will decline.

Critics from both sides will claim, at the same time, that “regulation is too high,” and “regulation is too low.”  They will both be describing aspects of the same proverbial elephant.

The ability of a new regulation to pass a (partial equilibrium) cost-benefit test does not mean said regulation is a good idea, at least not without adjusting for the “crowding out” effect.

Hiring more regulators will address the dilemma only temporarily (assuming that the number of regulators cannot keep up with the number of laws).  At first more regulations will be enforced, but as time passes the quality gap between the enforced regulations and the most important regulations will again open up.

If you think you just passed some really, especially important regulations, slow down the pace of future regulations.

If you are pro-deregulation, slowing down the pace of forthcoming regulations won’t much help your cause.  It will shift back attention and labor to slightly more important regulations, however.  It is possible that the net regulatory impact goes up.

Sunset provisions may induce regulators to treat, enforce, and monitor the old regulations more like the new ones, and vice versa.  What other institutions might contribute toward this end?


Interesting post... sounds like a problem right out of my Game Theory class...

I think the idea of "essential" and "non-essential" regulations is not so clear cut. An old regulation will start out as essential, but then the "regulatees" will find a way to circumvent it and engage in the behavior that is supposed to be limited by original regulation. Then a newer essential regulation is required to stop the new means of engaging in the illicit behavior. Innovations by the "regulatees" eventually lead to many regulations aimed at preventing the same behavior or outcome. Each regulation is necessary because the "regulatees" can always use whatever innovation is not being regulated to engage in the illicit behavior. In the end you get a skyrocketing number of regulations that are each necessary but not sufficient for achieving the regulatory goal.

Tyler, you have been posting the "science" of macroeconomics (posts about IS/LM models) and the "art" of policymaking (many posts on monetary and fiscal policies but also some about regulation of all sorts of human actions). Today the WP publishes these two pieces --one about "science", the other about "art" of public policy-- that make clear the stupidity of claiming that any "scientific knowledge" is settled and how limited the role of "scientific knowledge" in policymaking is:



I hope you always remember those two points when posting about "scientific knowledge" --and so your readers when they comment on your posts.

Well, except for AGW and AD.

Regarding bank regulations, I think the problem is that political exigencies may distort economic solutions.

Plus, if the supposed cause of the problem is not the true cause of the problem . . .

I'm fond of Heinleins idea from The Moon is a Harsh Mistress:

Divide your congress randomly into two chambers.

The sole power of chamber 1 is to pass new law. Passing a new law, requires X% of the votes.

The sole power of chamber 2 is to remove existing law. Removing a law requires y% of the votes.

y is smaller than x.

The setup ensures that a class of politicians exist whose only power is the power to deregulate. They are thus forced to choose between having no influence, and having influence by removing bad law. The assumption is that they would respond by actually spending time looking for laws that needs removing - something that few politicians do today.

In the book x was 66 and y was 50. The reason there should be some difference, is to avoid flip-flop effects where chamber A votes in a new law which is then immediately removed by chamber B and so on. (i.e. it ensures stability of law)

Assuming the quantity of legislators in both chambers is equal, then y < x does not avoid 'flip-flop effects.' To prevent a recently passed law from being immediately removed would necessitate a rule governing the procedures between the two houses, i.e., a constitutional rule.

Another option for producing stability would be to go with what Centinel (Samuel Bryan), an Anti-Federalist from Pennsylvania during the debate over ratifying the Constitution, proposed in his first letter. He sought a built-in delay on laws taking effect so that the legislature would have time to retract any ill-conceived laws before they were able to unsettle society.

Agreed in parts; actually, each regulation/law should be derived from a fundamental law; that is the reason we have a constitution.
New regulations could be revisions/ reinterpretations of older laws; or in case they are entirely new, they should be in response to some radical event/ technology shift, in which case they are as important and timely as any age old law.
There is no reason for the over time average quality of regulation to decline; quality of regulations depends on quality of legislators and should be independent of temporal factors.

Doesn't technology create at least a mitigation of this effect? Automobiles allow officers to respond more quickly to calls while computers allow regulators to more quickly sift through the records of a company to tabulate any and all laws that they may have broken.

I do like Heinlein's idea in principle but think that it would require a sort of reverse sunset provision. That is, a law must stay in effect for X years before it is repealed so that we can see how it does.

"Reverse sunset provision"? I don't recall any such provision inTMiaHM. If a law is so idiotic that as soon as people see it, they realize it should be repealed, why wait? (I'm not mentioning any specific recent legislation here)

The problem I see with that analysis is the regulators don't just enforce regulations, they're also the ones who write them. Obviously, it's not necessarily the same individuals or job categories, but the regulations that are actually enforced are written by bureaucrats (presumably following the principles contained in laws). Which is why we see call for reform to prevent regulators from serving as unelected, de-facto legislators -- for example:


All of which changes your dynamic -- more regulators means more regulations, not just more enforcement capacity.

First, the author you refer to takes a selective view of the Constitution. Hammerton points to the opening clause of Article I of the Constitution, which vests authority over laws in Congress. However, he then ignores the meaning of Art. II, section 2, which states:

"He [the President] shall have Power, by and with the Advice and Consent of the Senate, to make Treaties, provided two thirds of the Senators present concur; and he shall nominate, and by and with the Advice and Consent of the Senate, shall appoint Ambassadors, other public Ministers and Consuls, Judges of the supreme Court, and all other Officers of the United States, whose Appointments are not herein otherwise provided for, and which shall be established by Law: but the Congress may by Law vest the Appointment of such inferior Officers, as they think proper, in the President alone, in the Courts of Law, or in the Heads of Departments."

The authority to appoint 'public Ministers and Consuls' as well as Congress' authority to vest power in 'inferior Officers' and 'Heads of Departments' with the President makes it very clear that the Constitution was drafted with the intent that subordinate agencies would be required. As to the specific power for agency rulemaking, the Supreme Court has reviews the scope of this authority repeatedly and while the Court does find specific rules in violation of the Constitution it has not issued a blanket statement denying Congress the authority to delegate responsibility for implementing or further specifying laws.

Second, your claim that "the regulators don’t just enforce regulations, they’re also the ones who write them" lacks specificity. To take one example, the EPA is tasked by Congress under the Clean Air Act to establish National Ambient Air Quality Standards to be reviewed every five years. Are you claiming that the scientific advisory committee that helps to determine a reasonable standard is composed of 'bureacrats'? What example(s) did you have in mind? The lack of specificity in your argument means that one must argue against your undefined generalities.

Finally, the Constitution intentionally put some aspects of the government out of the reach of popular elections (the Senate until the 17th Amendment and the Supreme Court today), that is, they are not directly accountable. Perhaps we should consider that democracy isn't an inherently good system of government. It is as good as its citizens. The framers of the Constitution took into account that a system of mixed government may offset some of the deficiencies of popular government.

I wasn't endorsing the linked article in whole-just in support of the claim (which should be pretty uncontroversial), that one of the main tasks of regulators is to write the detailed, legally-enforceable regulations authorized by new laws (and to review and update regulations for existing laws). It's also pretty clear that regulation is used as an end-round when legislation fails (as when net neutrality rules that failed in congress were 'passed' and imposed by regulators instead.

And a simple theory of regulatory uncertainty is that uncertainty rises with the disparity between #regulations and #regulators, at least over parameter ranges where not everything can be enforced. Then a rise in #regulations or a fall in #regulators increases the uncertainty over which regulations will be enforced, and on whom they will be enforced.

Many laws are generation specific, and lose relevance over time. I think most law should "expire" and sunset by default or require renewal after X years. We have term limits on many politicians and lawmakers, so why not on many laws...

>We have term limits on many politicians and lawmakers

Yeah, but the not the politicians that we need it for the most.

This is an important point. It reminds me of an article I read about California's Department of Alcoholic Beverage Control. Basically, they found an old regulation in the books that prohibited "rectification" of distilled spirits. The purpose of such a regulation after the Prohibition era was obvious, it was trying to keep consumers safe from sketchy production methods. However when the ABC found this old regulation, they interpreted it to mean that they had to shut down any place that served infused liquor, a coconut-infused rum, for instance. Needless to say this was a colossally stupid thing to regulate, and state legislators had to waste their time repealing this old regulation. The point is this: history matters, human social patterns and practices transform, and therefore it isn't at all clear to me that old regulations will be superior to new ones. It is an attractive Burkean proposition, but one that probably deserves more scrutiny.

For reference: http://www.sfgate.com/cgi-bin/article.cgi?f=/c/a/2011/08/26/BAIE1KRJ20.DTL

"If the best or most fundamental regulations are the ones issued first..." I'm not pro regulation by any means, but this given is not necessarily true. Regulations are a reaction to societal and technological changes in the market place, so you have to distinguish between new regulations that are addressing a new product, market, issue, etc, with new regulations that are merely tweaking previous regulations that addressed some fundamental change in the market. (Not to mention those regulations that tweak the tweaks, and those that tweak the tweaks of the tweaks.)

A good example of a good first issued regulation that continues to matter is 10b-5 of the Securities Law which criminalizes securities fraud even in transactions that are not public offerings. It runs a sentence or two and still does the lion's share of the enforcement work.

There is a further constraint on the number of regulators. Like any organization, there is a diseconomy of scale, so one approaches some asymptotic limit as to how many regulations can be enforced. That is, adding new regulators after a point only succeeds in reducing the unemployment rate among regulators.

Moreover, there appears to be no consideration as to whether the regulations are covering mutually exclusive sets of events. Hence we get requirements to both do a particular thing, and at the same time, refrain from doing that thing. But more to the point, no one can possibly keep all the rules in mind. I submit that the IRS's tax code is at this level of complexity, and the financial laws are next. Now combined .....

Tax regs are another good example of regs that generate a flexible private sector pool of regulation enforcement officers via compliance professionals. The ratio of CPAs and tax lawyers in the private sector to IRS employees for regulation intense areas of the law is huge.

Like Steve, I'd like to know what appellation you'd give to Q. You start by defining Q and then never refer to it.

Output of regulation.

Your answer reminds me of Condorito --the main character of Chile's best-known comic strip. Now it's time for Daniel Klein to fall backwards with a PLOP!

On Condorito read

and those that can read Spanish may enjoy this strip (any similarity between Condorito's answer and Tyler's is purely coincidental):


Condorito is simply brilliant. I read him during maths classes at school in Santiago.

Output meaning something like "restrictiveness-this-period", yes?

Daniel, don't be cruel. He's just having a Qisis.

"There are always more laws than there are regulators to enforce them, and thus the number of regulators is the binding constraint. ... assuming that the number of regulators cannot keep up with the number of laws ..."
Those seems like important pieces of this model, and I don't think you've presented any argument or evidence why they should be true. Can't one regulator enforce many laws? Isn't technology improving their ability to do so?

The regulated also have the same technology to improve their ability to violate the regulations, so no.

Both Scotland and England have bodies whose job is to look through hundreds of years worth of old laws and recommend some for scrapping. They can't keep up with the rate at which the politicians pass new ones, of course.

Breaking the model by not making regulation entirely dependent on the regulator bottleneck. This is one reason I don't like victimless crimes. The citizenry is involved in enforcing legitimate regulations, and if they aren't, that's a clue that they aren't legitimate.

My father was a fire chief and he used to say that he could close any building in town for code violations. Law makers tend to make more laws in reaction to something that happens due to ignoring of the existing laws, instead of enforcing the existing laws. It seems no wants to enforce laws particularly Democrats though Democrats seem to want to make more laws than Republicans. I agree with Andrew when he says "citizenry is involved in enforcing legitimate regulations, and if they aren’t, that’s a clue that they aren't legitimate". If the typical citizen does not turn in a neighbor for using drugs then it probably cannot be against the law. If the worker does not tell on his employer's safety reg violations the regulation should go. I really have doubts about whether regulation accomplishes anything at all.

Building code regs aren't primarily enforced by enforcement actions brought by government regulators. They are mostly enforced by influencing how architects and contractors do their job and establishing customs and practices in the industry. New building materials are built to code, architects build rules of thumb on issues like bathroom or stairwell size or the desirability of steps v. ramps based on them. and so on. Enforcement actions mostly just tag recalcitrant laggards. Buliding codes also receive lots of private enforcement by setting standards of care in tort actions and in contract actions where the codes are incorporated by reference.

Marginal players may not comply, but big time players have so much risk for non-compliance that they have to play straight and the economy is dominated by big time players. Banks insist on certifications of compliance before they fund loans, and so on.

Building code compliance turns out to be a pretty good discriminant between first world and developing economies.

When a regulation is first passed, it requires firms to do new things--install seat belts, deploy new production technology. Eventually, adhering to the regulation become part of the firm's Standard Operating Procedure and less effort is needed by regulators to ensure compliance.

It is also the case that firms will often continue with existing processes as long as there is no perceived pressure for change. Environmental regulations can force the firm to reexamine processes that have become SOP and find new modes of production that are both environmentally less harmful and more cost effective at the same time--for example recovering and selling at a profit materials that used to be discharged as effluent. Again, less effort by regulators will be needed once the firm has discovered that adherence to the regulation is more profitable than their old way of doing things.

While it is not always the case that adhering to regulation turns out to be more profitable in the long run, when it does, there is less pressure on regulators and they can focus on those regulations where there is an ongoing tension between compliance and profitability.

It seems there are some corollary issues here. The need for new regulations also arises from the ability of people to circumvent the old regulations. Plus, it is easier to get around poorly written regulations. And, what about the complexity of regulations, contradictory regulations, etc . . . Not too mention who watches the watchers? We need regulations to keep the regulators in check.

Such a big mess . . .

Unbelievably cynical. And probably true.

If you install incentives for politicians, then, of course, this problem goes away.

But we're not allowed to talk about that, are we?

So, a commenter asked about PhD topics. Here's one: "What's a Politician Worth?" If we are speaking term limits, then clearly very little. Would you term limit your cardiologist? ("Oh, you can practice for four years, but then you'll be captured by the AMA, so you have to retire then.") Probably not. If you do it to a Congressman, then clearly expertise there has very little value. So governing the country is a no-brainer, by this line of thinking.

On the other hand, it would seem Singapore can deliver about the same services as the US for half the cost as a pct of GDP. And they, of course, pay bonuses to politicians and bureaucrats. So by that measure, the difference between good and bad governance the in US is probably $1 trn per year, certainly well into the hundreds of billions. Take, say, 2% of that (on $500 bn) and that's $10 bn for your bonus pool. Divide that by 535 members of Congress, and you get about $20 million per member per year. That's probably about right, but bonuses on the order of $2 million per year would probably do the trick. $10 per household per year. I would gladly pay 100 times that, and consider it cheap. I actually think the members are worth $200 million per year, each.

There's been a lot of work done on how the electoral process incentivizes 'problem solving' rather than 'problem preventing', as credit becomes an electoral commodity, so to speak.

Further, while the theory is nice, I’d like to see some data on regulatory enforcement over time. I suspect it would very highly on the type of regulation, costs of enforcement, etc… Off the top of my head, if enforcing the regulation actually brings in revenue (think speeding tickets), then I could see enforcement having no trouble keeping steady.

At the state level, this is the normal routine. Removing regulations has the disadvantage that it has to be done out in the open: statutes must be modified, hearings specific to the changes must be held, and so on. The budget process, on the other hand, is both ridiculously complex and enormously boring, so it is much easier to limit the resources available for inspection and enforcement than to get rid of the regulation itself.

The down side of this is, of course, the occasional scandal after people die because a regulation was not being enforced. In the last few years, several states have had children in foster care die because the state badly underfunded the inspection side of their foster care system.

Agree. The better solution might be to give more resources to the regulator to reduce regulation and conduct regulatory reform.

People need to learn that there isn't anything really underfunded. It's part of the total package.

Andrew, If that were true, you wouldn't see those who are regulated be the most strident on underfunding an agency that regulates them. They aren't stupid. They know that if an agency is underfunded, it benefits them.

I put this comment on Robin Hansen's blog awhile back regarding bank regulation, but I think the argument holds for more general types of regulation. And it counts as a counter-argument to the initial claim: the best or most fundamental regulations are oldest. My claim is that any regulation has unknown efficacy. Older ones will just seem better because they have survived evasion, lack of enforcement and elimination -- but like in evolution, the oldest species are not the best, but simply the best adapted to the particular series of troubles they've seen. (Except sharks. They are the best.)

Begin quoting myself:

Let’s assume no one really knows what they are doing, so carefully crafted policy runs the gamut from good to bad.

When enacted, it is plausible that banks will first comply and then try to evade regulations. Compliance is expensive, but inventing a workaround is expensive and hard.

Under this model, regulatory workarounds that are worth more money to the bank, and thus to the consumer who chooses their services in the market, will eventually be used to evade regulations. Regulations that “work” will benefit everyone. You start with a spectrum of regulations from good to barely good to neutral to barely bad to bad. Bad regulations are evaded. Barely bad ones are not, but represent the cost of doing business for ignorant voters/consumers who would rather spend more time doing fun stuff than paying attention to bank loans or regulatory policy. The end product is mostly positive and the negative can be considered the cost of not having to engineer policy as well.

Also, from this model, it seems any process of deregulation would be bad. Like the regulators proposing a span of regulations from good to bad, deregulators would try to remove regulations that run the gamut from good to bad. Since bad regulations are evaded, their removal does nothing. However, good regulations are likely to be removed.

Again, since no one really knows what they are doing, business will go right ahead and try new things in the areas where the good regulations were removed.

And since no one really knows what they are doing, we never actually know which regulation in place is good or which deregulatory action led to a financial crisis.

I'm not sure that the assumptions that new regs are the ones that get enforced or that the older regs are more important than the new hold at all.

The counterargument would be that new regs aren't as widely disseminated and in any industry where self-regulation is the main component of compliance, less knowledge means less compliance.

A point related to, and building on, the one made by MAS. Some regulations, even ones fairly controversial at first, change society and become embedded social norms that won't need much or any enforcing after a while, thus freeing up space for new regulations. If we dropped all laws and regulations requiring restaurants not to discriminate by race, how many would go back to segregation? No smoking at work might be or become another social norm that would survive the lack of continuing regulatons (I would think the same effect is also in play at restaurants and bars, though that may be as much or more markets than just social norm changess).


Or, perhaps some regulations simply codify norms.

The binding constraint you identify is relevant to Police Patrol style regulation. (See McCubbins 84). Increasingly, regulation is implemented or executed using Fire Alarm style mechanisms.

Likewise, failure to enforce on-the-books regulations invites increasing intensity of litigation to compel enforcement (NRDC spends much of its budget thereon). On the whole, however, one could argue that court-enforcement imposes a cost to limit the number of attempts to enforce less meaningful regulation. However, private costs do not encompass public costs to enforce, so it's far from perfect.


But the Regulators conscript professionals who advise the regulated by promulgating regulations that inflict lots of pain unless those professionals advise in a manner that exceeds their professional obligations and trade group requirements. Check out Circular 230.

Lots of new regulations replace old ones or are incorporated into the way that industries do business and hence don't need much active enforcement. For example, lots of environmental regulations primarily influence what kinds of new capital equipment can and cannot be marketed and don't apply when no equipment is being bought and while sometimes costly doesn't take much active intervention by regulators in the affairs of private companies that hire their own compliance departments expanding the pool of regulators in effect.

Most regulations have trival effects limited to highly specifc industries. A few have very broad effects. A model that views all regulations as created equal misunderstands the nature of the beast.

This is such an abstract posting. Everyone has their own concept of "regulation", a certain "regulated industry", or a certain "regulator".

Unless you focus on something specific, this is just a post where you can identify your "type", and signal your set of beliefs without having to focus on something specific.

It would be more interesting, if you want to be general, if you focused on "types" of regulations--those which specify how things must be done or not done, or those which simply set objectives which the seller can meet any way they choose.

Or, if you had focused methods for cost/benefit analysis, valuation of life or injury, or the interplay between regulation as a standard of care which immunizes companies from further liability.

But, you didn't.

Interesting, but the problem with sunset laws is that they don't work, because they are not in the politicians' or bureaucrats' interests. See Chris Mooney, A Short History of Sunsets, Legal Affairs.

Are sunset laws in your interest?

Those laws against murder and theft have been around a long time and are pretty old, so it is long past time to get rid of them.

But the accompanying laws against adultery and idolatry have been eliminated by We the People electing people who erased them from our set of laws, much to the dismay of many conservatives who think those old regulations written on stone tablets must remain in force.

If the laws against adultery and idolatry were sunsetted, conservatives would have condemned sunset laws as a liberal plot to destroy America and civilization.

Would an Office of the Repealer work like they seem to be trying in Kansas?



A related issue to the dynamic of growing regulation is what I call rule wrestling. If rules or regulations are simple and fairly consistent, it behooves the players to operate within said rules. If, however, the rules are constantly changing or growing, the dynamic begins to shift from winning within the rules to winning via the rules. This can lead to regulatory arms races. The rules become more complex as players pass them for profit or even just as leverage against competitors. Resources are reapplied from producing value to fighting for value via the rules.

Comments for this post are closed