Tuesday, August 12, 2014

FDA and the costs of regulation

The Wall Street Journal has had two recent articles on the FDA, "Why your phone isn't as smart as it could be" by Scott Gottlieb and Coleen Klasmeier on how FDA regulation is stopping health apps on your iphone, and Alex Tabarrok's review of "Innovation breakdown," the sad story of MelaFind, a device that takes pictures of your skin and a computer then flags potential cancers. The FAA's ban on commercial use of drones is another good current example.

One of our constant debates is how much regulation or the threat of regulation is slowing economic growth.  Over the weekend, for example, Paul Krugman, finding the New York Times itself too soft on libertarians,

Actually, the cost of bureaucracy is in general vastly overestimated. Compensation of workers accounts for only around 6 percent of non defense federal spending, and only a fraction of that compensation goes to people you could reasonably call bureaucrats. 
And what Konczal says about welfare is also true, although harder to quantify, for regulation. For sure there are wasteful and unnecessary government regulations — but not nearly as many as libertarians want to believe. When, for example, meddling bureaucrats tell you what you can and can’t have in your dishwashing detergent, it turns out that there’s a very good reason. America in 2014 is not India under the License Raj. 
Well, maybe, maybe not. Nothing in the FDA or FAA articles mentions the cost of the bureaucrats' salaries as the drag on growth, so that's a classic red herring.

The cost of regulations is the new businesses that don't get started -- or that fail as MelaFind nearly did, because the Raj would not grant a license -- the innovative products they would bring us, the employees they would hire, and so on.

The trouble is, these costs are awfully hard to measure. In the big demand vs. distortions debate (e.g. here) for our current stagnation, how do you put numbers to anecdotes such as these, and then add them up over the whole economy?  So far, it hasn't been done. Like the Laffer curve, we sort of know where the end point is -- even Krugman understands the stagnation of the License Raj, and Hernando De Soto is pretty convincing on regulatory-induced stagnation in other countries. But where are we on that spectrum? Krugman has no evidence that it's small. And I have no solid, quantifiable evidence that it's big. How do we do a "Potential GDP" that adds up these costs, as the CBO attempts to add up capital and people?

For the problem is not really in the cost of the regulation as written down. The problem is the cost of the regulatory system, the cost of the whimsical, political, and discretionary actions taken by regulators, as in this instance.

An important lesson in economics, and in science is, that which you can't measure you tend to ignore. So it's quite easy for us to go on, with economists such as myself reading these anecdotes and inferring we have a major problem, and others convinced that there's nothing here that a trillion dollars or so of "demand" wouldn't cure. Finding a way, even a conceptual framework, to add up these anecdotes would be a big breakthrough.

It matters for the big macro debate. It also matters as we think about financial regulation.

I can't resist a late snarky note on my dialog a few years ago with Glen Weyl and Eric Posner over their proposal that the government create an FDA-like agency to evaluate all financial products before the government allows companies to market them.  I wonder, after more and more stories like this come out if they still think it's such a great idea!

Prodded by Glen and Eric, I've been struggling with the idea of cost-benefit analysis for financial regulation. We need some sort of structure to evaluate all the clever proposals agencies are unleashing un us. But how do you add up these kinds of costs? And not end up, like Krugman, counting paperwork hours and bureaucrat salaries that are specks on the tip of the iceberg of the real costs?


  1. Looking back further in my Feedly, it looks like Casey Mulligan talked about this last year in a series of posts and laid out some of the data issues, as well as recent research on some possible empirical solutions. I found them to be a nice refresher on this topic.


  2. If you believe that the costs of doing business measures are a proxy for socially-inefficient regulation, then there is a ranking by the Doing Business project at the World Bank.
    The ranking is a measure of business regulation. And the US is doing pretty well relatively to other countries.

    I think the hardest question is how to differentiate socially-efficient and socially-inefficient regulation. Is forbidding commercial use of drones socially efficient or not? I think there are arguments both ways. The same for the FDA ban on the mobile app. Should we allow the app to make diagnostics and maybe (because of a false-negative) not look for a doctor when you should? Is the app "practicing" medicine without a license?!

    1. Great comment. I need to review the world bank methodology in detail.

      I picked the FDA example since it seemed so obviously over the top on socially inefficient. It's just a camera that looks at your moles for goodness sake. You can look at your moles and wikipedia or web md -- and we all do that. Next thing you know they'll ban google searches for "does this mole look bad" and insist all such pages say "information not available by order of the federal government, go see your licensed dermatologist." There is a deeper point; web and apps are making do it your self medicine more and more important, just like do it yourself everything else. And guess who is in the way.

    2. "Next thing you know they'll ban google searches for 'does this mole look bad' and insist all such pages say 'information not available by order of the federal government, go see your licensed dermatologist.'

      Yes! Exactly! Excellent analogy!

  3. The GAO said in 1999, that of 101 “economically significant regulations issued by the EPA from 1981 through 1998, only five were the subject of retrospective studies.” So they are not even checking their direct cost-benefit estimates (as opposed to the more difficult general equilibrium effects you bring up).


  4. >Nothing in the FDA or FAA articles mentions the cost of the bureaucrats' salaries as the drag on growth, so that's a classic red herring.

    Except Krugman was talking about:

    >Mike Konczal takes on a specific example: the currently trendy idea among libertarians that we can make things much better by replacing the welfare state with a basic guaranteed income.

    For which the admin overheads of UI and SS is very relevant, not a red herring at all.

  5. "The FAA's ban on commercial use of drones is another good current example."


    Agree with rest of the article but not this! Allowing widespread use of drones in populated areas is extremely dangerous -- for a reason which nobody (as far as I know) is talking about: Viruses!

    It would be extremely easy for hackers to have undetected virus software on a a whole bunch of drones, and then take over all the drones in a certain area and have them immediately fly to -- and crash at high speed into -- a single designated target (such as, say, John Cochrane walking down the street at lunchtime).

    Being hit by one such drone might (might!) not do serious permanent damage. Being hit roughly simultaneously by a dozen (or more), mostly in the head (you won't be able to block more than one or two with your arms) easily could -- in fact, it would probably kill you.

    Then what do your heirs do? Sue the drone manufacturer? Yeah, that's like trying to sue Microsoft because their shitty software is so prone to viruses.

    As personal computers have shown, no matter how much precaution you take some people's computers will always be infected and important data destroyed. But at least the desktop virus can't physically injure or kill you. But a group of drones can.

    And, of course, the NSA will probably have a backdoor into the drone software as well. Do you really want dozens of NSA-compromised drones buzzing about as you walk back from lunch, ready to write your next piece documenting the latest destructive government meddlings in the economy?

  6. To give an idea of the scale of the costs, we did a study on privacy costs in Canada. The bureaucrats costs only $40 million per annum...the private sector privacy offices which they legislate only cost another $70 million. But the administration costs are more like $3.8 billion and the changes in behaviour reduce investment by $1.8 billion per annum. So that gives you an idea of how important the bureaucrats' salaries are in the scheme of things...we called our study Exploring the Iceberg for good reason!

    1. Try www.conferenceboard.ca

      You can find the study by searching eLibrary under Exploring the Iceberg or by me as author...Michael Grant


Comments are welcome. Keep it short, polite, and on topic.

Thanks to a few abusers I am now moderating comments. I welcome thoughtful disagreement. I will block comments with insulting or abusive language. I'm also blocking totally inane comments. Try to make some sense. I am much more likely to allow critical comments if you have the honesty and courage to use your real name.