Federal Regulation Is Not the Cause of Declining Dynamism

My paper with the excellent Nathan Goldschlag, Is regulation to blame for the decline in American entrepreneurship? has finally been published. Our paper tests the plausible theory that regulation reduces dynamism as it builds up over time. Michael Mandel explains:

…it’s possible for every individual regulation to pass a cost-benefit test, while
the total accumulation of regulation creates a heavy burden on Americans. The number of
regulations matter, even if individually all are worthwhile.

I call this the ‘pebble in the stream’ effect. Thrown one pebble in the stream, nothing happens.
Throw two pebbles in the stream, nothing happens. Throw one hundred pebbles in the stream,
and you have dammed up the stream. Which pebble did the damage? It’s not any single pebble,
it’s the accumulation.

This is also the theory of regulation and declining dynamism that Mancur Olson puts forward in his classic, The Rise and Decline of Nations. We find, however, that declining dynamism cannot be explained by growing federal regulation. The reason turns out to be simple: the decline in dynamism is widespread across many different industries and, in particular, it is widespread across heavily and lightly regulated industries. Our finding does not imply that regulation is necessarily good–regulations could fail a cost-benefit test and yet not have much of an effect on dynamism–nor does it imply that no regulation could explain declining dynamism only that we should probably look elsewhere for an explanation of declining dynamism than the cumulative growth of federal regulation. See the paper for some suggestions.

Frankly, it’s difficult to publish a paper that fails to reject the null hypothesis. A positive or negative effect is a natural stopping point–ok, they got it, let’s move on–but a zero-effect always leads to complaints that you didn’t run the regression in such and such a way or you could have done such and such a test. The asymmetry in paper evaluation leads to the file drawer problem where published results tend to reject the null even when a random sample of all results would find that the null is supported. We know the file drawer problem is serious because it predicts that studies with small sample sizes should have larger effect sizes–an effect that has often been found.

I can’t complain too much, however, because our paper was published in Economic Policy, a highly-ranked journal, and is the Editor’s Choice paper for that issue. The referees certainly made the paper better.

One of the things we did in the paper to counter the claim that our methods or data were defective was to look for entirely independent tests of the regulation hypothesis. If regulation is the main cause of declining U.S. dynamism, for example, then we ought to find that declining dynamism is associated with declining industry size. But when we look at dynamism, as measured by excess job reallocation rates, and industry employment what we see is that dynamism is declining in both shrinking and growing industries (see above). The paper has many additional tests.

The data and tools in our paper have other applications. Our methods, for example, can be used to distinguish between special-interest and general-interest regulation and could be used to test many other theories in political economy.


The reason turns out to be simple: the decline in dynamism is widespread across many different industries and, in particular, it is widespread across heavily and lightly regulated industries.

Doesn't that assume that these industries are independent? That regulating one does not affect the others?

Is that justifiable? An industry might be lightly regulated but if it has an impact on the water courses nearby or if it required buying in from a more heavily regulated industry like power then it will have trouble surmounting the regulatory burden on their suppliers.

Good point. Another problem/challenge presented by AlexT's paper is that conclusions are drawn from previous papers, like in a chain, so the weakest paper in the chain breaks the final conclusion. Case in point: "dynamism" is equated with startup/exit rates, and the age of businesses, yet "Perhaps less surprising, higher productivity firms are more likely to survive" and then "Dynamism, for example, could be relabelled ‘churn’ and reduced churn could be driven by better job matching and reduced uncertainty leading to a desirable consequence of longer job tenure".

So effectively: dynamism is both good and bad, depending on your point of view, ergo regulation could be good or bad. The paper should have pointed out what Hayek said, in that regulation could be good but that's irrelevant if you believe in the First and Second theorems of Economics (https://en.wikipedia.org/wiki/Fundamental_theorems_of_welfare_economics). No matter how good a bureaucrat is, what he or she can achieve could be done automatically under the free market; conversely, all it takes is one bad bureaucrat to undo what the free market could achieve.

Finally, the citation to Mancur Olson's classic "Rise and Decline of Nations" is interesting since that book makes a positive reference to Stalin, saying in a command and control economy such as the USSR you need to have a man like Stalin in charge, to reallocate people by fiat, since a weaker hand will not allow the system to survive. True enough, after Stalin made way to Khrushchev, Brezhnev and lesser leaders, ending with Gorbachev, the USSR collapsed (and good riddance, but it proved Olsen was right).

"A variety of explanations for the {dynamism} decline have been suggested... This research uses ... federal regulations {to} explain the severity of the decline in dynamism at the industry level. We find no significant relationship between federal regulation and changing economic dynamism. "


So ... although there are many possible reasons for the dynamism decline -- this research study somehow chose only one possible reason (industrial Federal Regulation) and ignored all the rest.

Thus, the very premise of this study immediately fails the common sense test.
Why was 'Federal Regulation of industry' singled out for analysis, in isolation from all other causes?

The narrow study's methodology and analysis are also very weak, it certainly does not prove that "Federal Regulation Is Not {a} Cause of Declining Dynamism".
The whole effort just seems a typical contrived academic publication push.

@AD2 - your criticism is unsound. The AlexT test is precisely to ignore all other possible reasons for the dynamism decline, and statistically test whether the dummy variable of "Federal Regulation of Industry" affects dynamism. It is found that this dummy variable does not, so the smart answer is that you fail to reject the null hypothesis, meaning you cannot prove at the 95% confidence level that the dummy variable of 'bad regulation' is responsible for dynamism decline.

One must "isolate" the specific variable under study to measure its effect or non-effect.... in a system of multiple variables. That was not attempted here, nor (apparently) even considered.

@AD2 - you confuse and conflate two different things: finding the cause(s) among multiple variables and rejecting the null. Granted, it's theoretically possible that several causes will work together to 'cancel out' each other, but that's rare. Here, AlexT is testing the null hypothesis that government regulation is the 'cause' of the decline of 'dynamism' and finds that's not statistically true, with 95% confidence.

So Much For Subtely's point about regulation having possible upstream and downstream effects is a good one and we use input-output tables to test for precisely such effects.

Alex writes:

Frankly, it’s difficult to publish a paper that fails to reject the null hypothesis. A positive or negative effect is a natural stopping point–ok, they got it, let’s move on–but a zero-effect always leads to complaints that you didn’t run the regression in such and such a way or you could have done such and such a test.

I would argue that there aren't anywhere enough papers published with results that failed to reject their null hypotheses. Where real science is concerned, finding out what *doesn't* explain an observation or predict an outcome can often be as valuable finding out what actually does. At the very least, the authors of the non-null hypothesis rejecting results are saving both themselves and other researchers a considerable amount of time and effort in getting to a better understanding as they rule out approaches that don't lead to definitive results.

Unfortunately, it can also be argued that the current academic environment, where non-null results are seemingly favored for publication to the exclusion of non-glamorous findings, is contributing to a lot of inefficiency in research. Not to mention an abundance of published results that unnecessarily steer follow-on research into dead end routes because they proved to be based upon the equivalent of false positive results, where the authors stopped because they got to something they thought they could get published rather than seeing their study through to confirm their apparent findings would hold up under stronger scrutiny.

Speaking of the file drawer problem....


Thank you for the link.

Dynamism = entrepreneurship?

What about innovation? Efficiency?

Well, whatever it is, dynamism hasn't stopped the rich from getting richer. A process that has seemingly accelerated even as dynamism has seeming declined.

Not that one should expect a paper from Prof. Tabarrok concerning that subject any time soon.

Wouldn't decreased dynamism tend to concentrate wealth, since existing companies would face less competition from innovative and cost-effective competitors?

@Alex - Were you disappointed in the results of your research?

Also, was it published, and got Editor’s Choice, because they were happy with the result?

Bankers look at risks when deciding their exposures and risk premiums... their yields.
But then came bank regulators and decided the the capital requirements should also be based on perceived risks; which essentially means that the banking system doubled down on perceived risks; which, since risk taking is the oxygen of development, guarantees a huge decrease in economic dynamism.


You wrote a paper claiming that Federal regulation isn't so bad, and you were worried that it would NOT get published?

That is one of the least-likely things I've ever read here. It is absurd.

Well, he is the Bartley J. Madden Chair in Economics at the Mercatus Center, and the odds of the Mercatus Center ever publishing any sort of paper that does not show problems with federal regulation are minimal, to put it mildly.

You really have no idea what sort of department Prof. Tabarrok is a member, do you? He is not surrounded by people with lifetime guaranteed employment funded by the taxpayers of the Commonwealth of Virginia who believe federal regulation is desirable, after all.

And yet he did publish it. Your deep understanding of the department may be flawed.

Related to this idea of broad loss of dynamism, I came across a rather pessimistic essay recently. I certainly don't buy into it fully. I still feel like an optimist, but I have to admit that the author has found some connections.

While I don't believe in American collapse, it is undeniable that we don't put that much energy into things.

Why We’re Underestimating American Collapse

We have all these school shootings, but what can we do? (a lot, actually) We have all these homeless, but what can we do? (a lot, actually) We have this terrible opioid epidemic, but what can we do? (a lot, actually)

You can probably think of a few more, and while you might not put "we elected this idiot president" in the list, I certainly would.

Americans have stopped being doers, and have become sufferers.

"We have all these school shootings, but what can we do? (a lot, actually) We have all these homeless, but what can we do? (a lot, actually) We have this terrible opioid epidemic, but what can we do? (a lot, actually)"

Feel free to do something. There are plenty of charitable Christian organizations that are actively working with the homeless and with addicts. You could start by donating some money and more importantly your time.

On the one hand, on the micro level, that is true. And I am sure many of us have helped or contributed at such local efforts.

But remember, it isn't really a solution unless it works.

Fully. On the macro level.

To be a bit harsher about it, "so go donate $5" is more an example of the surrender than the solution.

This isn't even complacency (which implies satisfaction), this is resignation.

The point is what has caused these problems and how can that be ended or reversed.

Giving to a charity that supports or subsidies the homeless may make the problem worse.

I am curious about relationships between industries that are heavily regulated and relatively unburdened industries that rely/interlink with heavily-regulated industries. I’m thinking of manufacturing and it’s relationship to raw materials (mining, fertilizer inputs, lumber, water, etc.), for one. There might be others. Am I accurately interpreting your definition of Industry?

Agreed. If heavily regulated product x doesn't get produced, that affects warehousing and many other lightly regulated industries. How do you tease out all the connections?

Anonymous, we use input-output table to look for such upstream and downstream effects and it doesn't change the results of the analysis.

There seems to be a disconnect between the America that is risk averse (thus, all the safety regulations) and the America that is risk loving (thus, Bitcoin). While business startups are at a historically low level, investment motivated by a belief in rising asset prices is at a historically high level. Maybe Americans and the American economy are highly dynamic, as measured by the willingness to invest in assets based on the belief of rising asset prices. Or maybe Americans aren't very good at evaluating risk? I don't invest in Bitcoin because I believe there's way too much risk. On the other hand, I regularly take long (25 miles) bicycle rides on the road. My grandmother took risks (planting a citrus grove, building what was known then as a motor court, etc.), but she would be mortified if she saw me riding my bicycle.

If you were worried about the null you should have modeled it with Bayesian, then you'd have the entire probability distribution to interpret instead of just the average effect.

"The reason turns out to be simple: the decline in dynamism is widespread across many different industries and, in particular, it is widespread across heavily and lightly regulated industries". That turns on how the authors measured the burden of regulation. It can't be as simple as counting the pebbles, because that would mean assuming that each pebble causes a similar impairment of enterprise. But, by God, I look at the paper and find that it is based on a pebble-counting method. "RegData counts the number of restrictive words or phrases such as ‘shall’, ‘must’, ‘may not’, ‘prohibited’, and ‘required’ in each section of text."

Heh. But they had to do that because it is impossible to read all the regulations and understand their import. If you actually did that, understood the import, the costs involved, and more importantly how people who are choosing whether to do something, you would be done three generations from now.

Then pronounce with certainty that it has no effect at all because we couldn't measure it.

They Have to do it this way because of Derek's point. But what it means is that the key right side variable is measured with considerable error. Classic errors in parables needs to diminution bias, bias toward zero. So that is to make it more difficult to reject the null. Mus

This method will result in lots of errors because the regulations are written different styles:

Large mandatory pronouncement, followed by multiple conditions, treated as less rigorous than a regulation that uses mandatory language for each individual condition.

Use of mandatory words in the exceptions or alternatives, which are apt to describe regulatory flexibility or alternative availabilities not present in other regulatory systems.

The use of the word "may" in a mandatory context will be overlooked.

What if "may" was used as part of "may not"?

Every industry is heavily regulated. Presumably they have an office somewhere, there are heavy regulations about every aspect and detail in the building and furnishings. Presumably there are employees. If you employ someone you are heavily regulated, expected to know and control every substance they come in contact with, the labor regulations themselves, etc. Presumably they plug some device into the wall to make it work. Another interface to a heavily regulated activity.

The problem isn't the regulations, for the most part they are about documenting good practice. The problem is that you cannot know the regulations that apply to your business. It is impossible, there are too many, and the precise meaning of them requires expensive legal proceedings. And every business person knows of someone who for some reason got the attention of the regulators who then proceeded to make their life a living hell.

It is called regulatory risk. The inspector changes, a new administration, and you risk losing everything. A legal challenge, even if you win, will cost you your profits and even your business. The regulators use that as a bludgeon.

And with an open global economy, there are places you don't face those risks.

If you want to really know the effect of regulation, do a study on compliance. You will find shockingly low levels of compliance and an equivalent hesitation to invest. There are better margins selling meth with less risk.

That last statement is not a joke. Where I live if you want to be free of regulation and taxation, with less risk, grow and sell marijuana. You may get a fine once in a while, but it is less than most businesses pay per month when they remit GST.

"we ought to find that declining dynamism is associated with declining industry size."



Is the field of economics unfamiliary with the concept of regulatory capture?

Consider employer obligations under the Patient Protection and Affordable Care Act. Protects established businesses, inhibits small business development.

+1, the impacts I've seen regarding the building industry impact the smaller companies that can't afford dedicated company lawyers. It's extremely expensive (on a project percentage basis) to get the proper Federal legal paperwork to demo an older industrial building. Then it's far more expensive to treat substantial portions of the demo as requiring HazMat handling.

At this point, it's best just to leave the buildings boarded up (unless it's prime land). Demoing the building is more costly than the property is worth. Think about that for a moment. The current regulatory structure encourages ignoring issues instead of fixing them.

I call them the Full Chinese Employment Act.

You have any references on that, I am interested in redevelopment generally and I'd like to look at this issue.

"You have any references on that, I am interested in redevelopment generally and I’d like to look at this issue."

Which part? I know a person who got very personal with the Feds over asbestos removal without proper Federal permits. Apparently, a) state permits don't mean anything to the Feds, b) the state doesn't actually know what the Federal requirements are and c) the Feds expect everyone to know precisely the current state of regulation and ignorance is no excuse. d) The Feds will start with a site Stop work order, which shuts down everything, not just the asbestos removal.

Do not, I repeat Do Not! get involved with demolishing an old industrial facility that has asbestos in it (hint: anything before 1974) unless you are absolutely sure that a specialized lawyer has signed off on the necessary permitting.

To give you an idea of the likely costs: "Removing a 10-foot section of asbestos pipe insulation could be $400-$650. "

I do wonder about this. From first hand observation, I know that EPA regulations (particularly asbestos removal & handling and 'wetland' protection) have significantly impacted the building industry.

In addition, it's irrefutable that tighter pollution regulations, significantly increased the cost of running a coal plant.

Definitely. Like a tax, however, regulation could make the industry smaller without making the industry less dynamic over time in terms of startups and job reallocation rates.

Alex T, thanks for the response. That would explain my observations and reading.

Indeed, the primary responses I've observed have been to Not perform upgrades (to smoke stacks or coal plants) and to Not demolish or refurbish older buildings (that almost universally contain asbestos) and to Not perform upgrades to facilities that would change their grandfather status with respect to 'wetlands'.

Intuitively, I would consider a smaller industry that didn't do as much to be less dynamic.

@AlexT - why are startups so important? Some mythology from the Clayton M. Christensen‎ worldview, or maybe extrapolating Intel into every other industry, or do you not realize that startups are essentially theft of intellectual property from the mother (incubator) company? As for job reallocation rates, your own paper rebuts this, correctly pointing out that you can have lower job reallocation rates and higher dynamism, to wit (from your paper) "Dynamism, for example, could be relabelled ‘churn’ and reduced churn could be driven by better job matching and reduced uncertainty leading to a desirable consequence of longer job tenure".

Bonus trivia: in terms of patents, most real innovation--pace Clayton M. Christensen type b-school book anecdotes--is done by big Fortune 500 companies. Mom and pop startups are often essentially stealing the big company IP and trying to make a go for it, hoping to be bought up later by the big company, and/or make enough money 'under the radar' to not really matter to the big company.

Alex: Nice work. Do you plan to follow up with an international study?


(But don't even think of touching Japan. That is *my* country to analyze. All mine!)

Restaurants (food) is highly regulated, yet we don't see a lack of dynamism in the restaurant industry. On the other hand, retail is lightly regulated, yet retail has almost no dynamism. Speaking of immigrants, if not for them Cowen would have to eat at, heaven forbid, Red Lobster or Outback. Speaking of natives, without the retail option what can they do to start a new business; they can't cook so a restaurant is out of the question.

Really glad to see this type of intellectual honesty, Alex. My priors, and almost certainly yours too I would guess, would have been for a positive result.

Curious if you preregistered your analyses/regressions too.

I'm curious about your definition of "economic dynamism." Scott A. Shane argues in "The Illusions of Entrepreneurship" that the go-to statistics on entrepreneurship are completely dominated by things like laundromats, taco trucks, lawnmowing services, and the like, and don't do a good job reflecting the kinds of innovative, technology-driven startups that we'd tend to associate with "economic dynamism."

The kind of economic dynamism I appreciate is the Vietnamese Pho restaurant that opened down the street last year. I am also impressed by the efficiency of my lawn service: the entrepreneur, an African American, and his two Mexican employees can service my large yard in less than 30 minutes, without them even talking to each other they know their jobs so well. And my contractor friend, don't get me started. He and his son can do about anything, and do it very well and efficiently. They have what I call useful skills. Flying cars and spaceships to Mars don't have much appeal to me.

Is "dynamism" what you're looking for, though? Is a high rate of entrance and exit a good thing in the pho market?

Really, I'm just looking for an efficient underclass to cheaply and quietly provide the kind of high customer service I desire. Everyone should know their place.

When it comes the growing consolidation and less small business, I always return to the simple belief that modern companies know how to take advantage of Economies of Scale than ever. And with growing technology the marginal cost of a byte is very small so companies take a lot longer to reach the increasing Marginal Cost and especially increasing Average Cost points on the curve. For all the hatred against the Big 4 - 6 banks, their sizes are growing the last 8 years. (And only rebuttal to this was Dodd-Frank reserve ratios were too high for small banks)

You have any references on that, I am interested in redevelopment generally and I'd like to look at this issue.

Measuring the effects of regulation also depends on what you measure as an effect as well.

Reducing the costs of an externality on others is not incorporated in the model since the only measure is the effect on an industry.

Thus, your findings may be even stronger: not only is there no effect on the dynamism of an industry, but, since you have failed to measure externalities and the reduction of externalities the case is even stronger that regulations may be more helpful on balance, and in no case harmful. As for dynamism, the paper also excludes the creation of new industries that may evolve from regulation, such as, for example, solar technology or more efficient consumption of natural resources with changes to equipment such as boilers.

I wonder how much of an effect mergers and acquisitions have? I worked for a very dynamic small software company that was purchased by a very large one. This large company proceeded to centralize all decision-making into ‘centers of excellence’ and took control of products away from the engineers who actually built them. I watched a vibrant, dynamic office where everyone had skin in the game and where all good ideas were listened to, turn into an office full of drones working on remote stranger’s often terrible ideas. Morale plummeted, and creativity along with it.

I would also look at changes to HR and the way dynamic and creative people are being rewarded. Our very large company dropped its employee evaluations because, to quote the HR person who trained us on the new system, ‘millenials won’t tolerate being criticised or ranked’. So we went to a ‘communitarian’ approach where we were supposed to encourage each other to better in a positive, inclusive way. Criticism not allowed.

The company also announced an official HR goal of 50% women in new engineering hires, despite women making up only 20-25% of new engineering grads. This has to mean we are picking employees who otherwise would not be hired.

Our large company is now in deep trouble.

My guess is that there is no single cause for a decline in dynamism, or even a handful of causes. It’s merely the way that characteristic of our economy has moved due to a host of changes.

"The company also announced an official HR goal of 50% women in new engineering hires, despite women making up only 20-25% of new engineering grads"

Yes, this is trendy in large companies. It also has the perverse incentives of ensuring that medium to small companies have an even more lopsided ratio of men to women.

It is obvious that Sarbanes Oxley has had a dramatic and obvious effect on Silicon Valley. Before Sarbanes Oxley, entrepreneurs hoped to go public and have an initial stock offering. After Sarbanes Oxley, they hoped to sell their company to Google.

This regulation has dramatically diminished the dynamism of the most dynamic sector of the economy, but this study does not detect that effect, and indeed could not detect it.

Comments for this post are closed