Patents Do Not Increase Productivity

James Pethokoukis points us to this graph in the CBO’s new report on innovation. Patenting is way up but total factor productivity has barely budged. 112414patent

Boldrin, Allamand, Levine, and Ornaghi show something similar cross-sectionally, namely that the industries with the most patents are not the industries with the biggest increases in labor productivity. At best there is only a slight positive relationship between patents and labor productivity (see the paper for more).




Is the productivity of patent laweyrs included in these numbers?

The second graph says it drops the "top 5%" (presumably of the income distribution). I guess that is necessary in order to get meaningful data, because otherwise you are counting the earnings people who benefit from patent-induced monopoly rents. But it might also introduce an anti-patent bias: if the newly patented ideas make a certain class of workers much more productive then those workers will be at the higher end of the income scale.

Ah, "industries with the most patents are not industries with the biggest increases in labor productivity"

Think about that for a moment.

Think about a mature industry--steel, flour milling, sheet metal fabrication, food processing--and think about an entirely new industry--chip fabrication machinery using new lithography technology, glass manufacturing (gorilla glass), 3d printer manufacturing, new materials manufacturing (nano particles, etc.)

Now, ask the following questions:

1. Which group of industries would you expect to see EVEN MORE patents in the future?

2. Which group of industries would you expect to see low productivity UNTIL the bugs are worked out from the new technology, while EVEN more new technology is also developing in the field and THAT technology is also simultaneously being added to the new technology.

Go back again and read the sentence: "industries with the most patents are not industries with the biggest increases in labor productivity." You may have read that first sentence to mean that patents do not lead to increased productivity, but when you understand that patents, and increasing numbers of patents, mean that you have to work out the bugs, that there are more inventions going on in the same field at the same time, that those inventions need to be get a different picture.

So, ask yourself: if there were a large number of patents associated with the introduction of the transistor back in the early 50's, and the process of discovery and adaptation was slow, did the transistor increase or lower productivity.

Taking into account The Great Stagnation, one might expect the amount of profitable patents to drop over time regardless of the productivity enablement of patents.

Yes, a time series is the wrong approach to analyzing this. A better approach would be to contrast and compare similar countries of varying patent regimes.

Or one could say productivity would have dropped without the additional patent activity.

So the CBO has the quite impressive ability to isolate the variable of 'patents' ... and accurately measure its effect on 'labor productivity' ?

How many variables are there in labor productivity ?

Wouldn't it be the case that patents in technological fields would lead to higher productivity in different (technological or non-technological) fields?

For example, if Amazon patents virtual server technology, it improves the productivity of system administrators (ie. enables organizations to fire system administrators to free them up to become baristas).

+1 The productivity increase in making a computer says nothing about the productivity effect of using a computer.

Why does the 2nd chart say "Top 5% dropped"?? Also, the first chart doesn't support the headline at all. First, uncorrelation is not uncausation. Second, hextupling of patent grant rates appears to have increased total factor productivity growth rates by ~25%. What's the argument against the interpetation that says we should just hextuple patent growth rates again and reap another 25% increase in TFP growth rates?

Excellent point, that I was about to make. Note the author James Pethokoukis (a Greek surname) is biased against patents, like AlexT, and his conclusions do not follow the data for several reasons. First, as Curt F. says, we can also read the data to mean "more patents would give us even more TFP". Second, in the Pethokoukis post, as commentator AIG points out, TFP does not include the stock of capital, which can be improved by patents, and anyway patents are simply the alternative, as AIG points out, to trade secrets, and not a magical potion to increase TFP. Historically, a big jump in patent filings occurred in the early 1980s when Reagan created a single new patent court to hear patent appeals, which increased the rate at which patents were upheld to roughly 66% from the below 50% rate, sometimes much less than 50%, from the 12 regional circuits that used to hear patent appeals.

In short, while I think the current system of patent reward is broken (for one thing, it does not reward the inventor, only the assignee, which is a classic agent-principal problem; for another, I do think we need different terms of protection for different industries, as AlexT agrees, with pharma needing more protection and software needing less), you could make the argument that having no patent protection would have resulted in even less TFP and productivity, which in any event slowed down after 1972 all over the world (even in countries with different or no patent protection).

It does not pay to be an inventor. One stat you should commit to memory: "The economist William Nordhaus has calculated that the inventors and entrepreneurs nowadays earn in profit only 2 percent of the social value of their inventions." - cited by Deirdre Nansen McCloskey in her recent Piketty review, citing Nordhaus, William D. 2004. “Schumpeterian in the American Economy: Theory and Measurement.” National Bureau of Economic Research Working Paper W10433

Does 2% capture sound fair? In short, inventors invent largely regardless of incentives, society benefits disproportionately, but currently patents do not give enough incentive to 'move the needle' for TFP. Abolishing patents however would probably make things even worse, at the margin, as Curt F speculates.

One would expect actual patents to reduce productivity somewhat for obvious reasons. It is the ability to get a patent that one hopes would increase productivity by spurring invention.

This is weak. correlation doesn't equal causation

And this isn't even direct correlation. Many patents will directly save money (materials) or increase the value of the product, but would not have any effect on labor productivity.

money equals labor, so saving money saves labor and thus increases productivity

higher value means it replaces more labor meaning increases productivity

Well, I hope I'm not the only one to find this completely believable. The costs of litigation have increased and the corresponding benefits to the plaintiffs have increased. This spurs both offensive and defensive patenting. The patents also get worse. I read patents daily (I'm a technical consultant). I've noticed in comparison to patents from previous decades each year has increased complexity, scope of claims, and an increase in obfuscation of disclosure. This isn't what people do to disclose innovations and receive protection as a routine business matter. This is a social cost imposed by how effective and expensive patent litigation has become.

I wonder how much of a cost the courts of East Texas impose on the rest of the United States. That would be a paper I'd like to see.

I share your belief that the modern patent system suffers from many problems, and that the quality of the "disclosures" in modern patent applications is lamentable.

What is unbelievable is that anyone thinks the data in the charts Alex blogged are convincing of anything, one way or the other.

Patent litigation isn't a "social cost". It's simply property right delineation.

That's like saying that putting up a fence around your house is a "social cost". Well, it is in the sense that it doesn't allow anyone from freely walking through your property. But, it's not really.

It's a property rights protection cost.

As products become more complex, and more dependent on other technologies, processes or products, then it becomes increasingly more important to delineate where...your property starts and where someone else's ends.

For the period 1963 to 1983, how can patent grants exceed patent applications? Sure, there can be patents granted for applications prior to 1963, but there will also be applications made up to 1983 that are not granted until after 1983. Not to mention some applications will be rejected without resulting in a grant.

@MT - it's because of the way the US Patent Office counts "applications". What they used to do more before 1983 is roll over patent applications several times before issuing them. So a single patent application would be refiled many times before finally being issued as a patent. This is a game patent examiners play, due to the way performance is measured at the Patent Office (which rewards rejecting patent applications initially, but then later rewards granting patent applications after they've been rejected a number of times). Also back in the days they had different laws which encouraged refling patents by both inventors and the US Patent Office, like 'submarine patents' such as Lemelson's troll patents, as well as the Gould laser patent (where Patent Office bureaucratic reluctance to issue Gould a patent ironically made him rich, as the laser market matured in the 20+ years it took to get a patent). Nowadays there's less incentive to delay by either side, since patents only get 20 years from filing, not 18 years from issuance, as before.

I'm not sure the evidence supports the statement. Other commenters have already pointed out all sorts of confounding factors.

Here's a gedanken: suppose we restart world history at the beginning of the Industrial Age, only this time, no one is allowed to patent anything. What is the effect on TFP growth?

Don't tell US patent lawyers (or their professional association) about this. They still believe there are BILLIONS AND BILLIONS OF DOLLARS at risk if we don't expand patents further and further.

This is not a recent problem. Consider why "Wright" hasn't been associated with aircraft manufacturing in a century. The original Wright Flyer used wing-warping, not ailerons, but the Wrights were allowed (wrongly) to extend their patents to cover them.

If they wanted a patent on the aileron, they should have built an airplane with ailerons and patented them instead of suing everybody in sight. To quote from Wikipedia:

"The Wrights' preoccupation with the legal issue stifled their work on new designs, and by 1911 Wright airplanes were considered inferior to those of European makers. Indeed, aviation development in the U.S. was suppressed to such an extent that when the U.S. entered World War I no acceptable American-designed airplanes were available, and U.S. forces were compelled to use French machines."

Enter FDR: "In 1917, the U.S. government, as a result of a recommendation of a committee formed by Franklin D. Roosevelt, then Assistant Secretary of the Navy, pressured the industry to form a cross-licensing organization (in other terms a patent pool)."

The irony is that the aileron had first been patented back in 1868. "If Boulton's ailerons had been revealed at the time of the Wright Brothers' patent filings, they might not have been able to claim priority of invention for lateral control of flying machines."

Base not case. You bring up one example, while I could bring up several counter examples: Tesla ripped off by Westinghouse and later by Marconi, Edwin Howard Armstrong ripped off by David Sarnoff, Dr. Damadian ripped off by GE (only redressed by litigation, Fonar v. General Electric, but denied co-sharing the Nobel Prize for the MRI as a final symbolic insult). Re your single data point, depending on how the Wright brothers patent claims read, the claims could cover ailerons, and a quick read of your very own Wikipedia cite shows that indeed the Wright patent apparently covers ailerons as well: "but the patent explicitly states that other methods instead of wing-warping could be used for adjusting the outer portions of a machine's wings to different angles on the right and left sides to achieve lateral (roll) control." OTHER METHODS. So the Wright patent was correctly upheld. Case closed, move along nothing novel nor unobvious here.

Cute but irrelevant.

1) There was prior art for the aileron.

2) Actually, that's beside the point too. But a big problem with patent law are IP conquistadors planting the flag on one small rock and claiming the entire continent of innovation for Spain.

3) David Sarnoff must have been an "artist" (according to Picasso) because he stole from everybody. But again, that is still irrelevant to the argument, which is about productivity.

4) Too bad Edison didn't bring the same open mind to Tesla's work that Faraday did to Maxwell's, with Faraday providing the experimental data and Maxwell doing the math. It helped that neither man had a financial interest.

5) As a result, Edison did great damage to Tesla and the fledgling power generation industry because he didn't have the mathematical background to understand what Tesla was doing and suffered from not-invented-here syndrome.

6) There are two sides to this coin: Yes, Sarnoff would have been better off (morally, too) licensing Farnsworth's work and not spending so much energy trying to stomp out FM the same way Edison tried to stomp out AC.

7) And the Wright Brothers would have been better off building better airplanes (their real genius was in aeronautical engineering) than engaging in a patent war, which was not at all a productive use of their brilliant minds.

8) Again, the question is not who did what first or who ripped off whom, but what is the most productive use of the inventor's time. "Because proprietary works for Apple" isn't a business plan.

9) IBM has played both sides of the coin. After its initial success with the PC, it wasted a lot of resources trying to create a proprietary PC platform and eventually threw in the towel. Now IBM is a big Linux shop.

10) Speaking of which, let's not forget SCO, the poster boy for suicidal IP litigation. It's still alive! Or is it? Hard to tell. Well, at least the lawyers are doing well (it'd be so sad if even they weren't).

Many good points, and I could rebut them, but like the proverbial tree that falls in the forest nobody hears, this is neither the place though perhaps the time to do so, as nobody besides you and I will likely read this, and maybe not even you. I do feel AlexT has good ideas about IP, like prizes, and the 'multi-tier' form of protection (e.g., software patents could use a fast track, like in Japan and Germany with their 'laid-open' 'unexamined' patents, that could expire in five years and only be examined if there is litigation, akin to copyright law) but he seems to enjoy bashing patents. Remember at heart all econ professors want you to work for no profit (MC = MR) with a horizontal AD curve. In short, they want you to be the proverbial wheat farmer found in Econ 101 textbooks. That's not a recipe for moving the Production Possibilities curve frontier forward. No small wonder Solow's growth model has technology as "exogenous", falling from the sky like manna! Goodbye.

There's no reason to link the two concepts. Some of the reasons have already been discussed above by other commentators.

Another reason is that patents reflect property right delineations...not necessarily "innovation".

A third...and perhaps MOST IMPORTANT that patenting activity really only took off in the 1980s. Prior to that, patents didn't have very strong property right protections, and hence most firms...DID NOT PATENT.

PS: Not to mention that "productivity" comes from process innovation, not "product innovation". Most patents reflect product innovation, and the most active product innovators aren't in fields where "productivity" is the main focus.

To those who claim physicists have it so easy compared to economists because economist can't run experiment in labs like physicists do, physicists use the earth, solar system, galaxy, and universe as labs to test theories, mining history for data to prove and disprove predictions based on hypothesis and currently operating theories.

Here we have an example of a hypothesis tested over the past century on how to boost various economic factors like growth, productivity indicating innovation, and so on.

I remember the arguments for a bunch of different patent reforms pushed by those I call free lunch economist who promised that better outcomes were possible by eliminating labor as the fundamental building block of the economy an growth and wealth creation.

By emphasizing patents, innovators would be able to create "value", growth, "wealth", etc faster without the slog of labor. Or government.

Of course patents depend on government picking winners and giving them monopoly power. whatever.

So, three to four decades later in this economist experiment and test of theory, the theory does not seem to have the benefits predicted.

You're not even remotely close.

First off, this is hardly an "experiment". This is James P. looking at two completely unrelated variables, at two completely incomparable time periods...and drawing completely unsupportable conclusions. (not terribly uncommon for James P., unfortunately)

Second off, "patents" aren't what people here think they are. Patents are "governments picking winners as losers" in the same sense that a government giving you a deed to your house, is "picking winners and losers".

The arguments explaining theoretically how state-enforced intellectual "property" cannot be property may be persuasive (and frankly correct). But that needn't matter to patent advocates. The advocates can continue to claim that patent enforcement stimulates innovation, whether patents protect property/natural rights or not. And although you hear from Mr. Everyman, as well as patent advocates, about how patents obviously create net incentive to innovate, patents also quite obviously have disincentives to innovate as well. So, the balance between the two really is the issue.

Are there any empirical efforts to try to quantify patent rights' relative incentives to disincentives to innovate?

Some substantial portion of the increase in patents granted is due to the need for defensive patents. Companies need their own patents for defense and countersuits if or when they get sued.

Comments for this post are closed