More explorations in stagnation
Suppose I am run a shop that builds engines and I figure out a better technique for inspecting my engines. I employ it and productivity goes up. My competitors hire away one of my engineers, find out the technique and also employ it. Pretty soon every engine manufacturer is using it and the world gets better engines. That learning by doing, knowledge spillovers and TFP all rolled into one. Its classic growth.
Now flash forward. Engines are being inspected by machine. I figure out a better program to inspect engines, I copyright it, sell it. That’s investment in equipment and software, profits for me as tech entrepreneur and capital deepening. Not so classic growth.
Well here is a chart of real GDP versus real investment in equipment and software.
Growth in investment in equipment and software is well outpacing GDP. At this point its still a pretty small fraction about 7.5%, however, we all know what happens when a portion of GDP grows faster than GDP, it eventually becomes GDP.
Here is the same thing in natural logs to give a better sense of the growth rates.
and once more on the same axis to show the gap closing
6 comments
Comments feed for this article
Saturday ~ June 11th, 2011 at 12:50 pm
Leigh Caldwell
Not quite sure what the implication is – your title suggests that IP protection of inventions might result in slower overall growth? Or do you mean that the productivity of the capital factor grows at the expense of the other factors? I guess the latter suggestion does fit with income patterns in the economy of the last 30 years.
Sunday ~ June 12th, 2011 at 12:57 pm
JW Mason
Interesting but I’d be careful about making too much of this. The growth of E&S investment relative to GDP is very sensitive to the different deflators used. This pattern is *entirely* the result of the specific hedonic pricing algorithm the BEA adopted for computers in the 1990s. (Note that in current dollars, the E&S share of GDP shows a strong downward trend.) Now, obviously the real price of computing power did fall, in a way not captured by the sticker price of computers. But the pattern you see here is going too be very sensitive to how we choose to measure that fall. Which means you probably shouldn’t put much weight on it.
Also, the combination of rising “real” expenditure and falling nominal expenditure suggests a low price elasticity of demand, which seems implausible for computers for a variety of reasons. So it’s more likely that your graphs are just reflecting a measurement error.
Sunday ~ June 12th, 2011 at 2:36 pm
JW Mason
Put it to this way:
Between 2000 and 2010, the price of computers relative to consumer prices as a whole fell by over 85%. Over the same period, the share of computers in nominal consumer expenditure rose from 0.65% to …0.69% (with a dip in between). And yet over that same decade there were plenty of factors — like the migration of lots of routine activity to the web, and of course simply higher incomes, since computers are clearly a superior good in the relevant range — that should have pushed computer demand up, independent of price.
So you’re really left with only two choices. Either computer demand is highly inelastic to price, or the BEA has gone overboard with its quality adjustments. The first does not fir my observation of the world around me; I doubt it does yours either. But if you take the second chocie, then the pattern in your graphs, goes away.
Sunday ~ June 12th, 2011 at 7:02 pm
Leigh Caldwell
JW: I don’t see how that follows: if prices have fallen 85% but nominal expenditure has more than kept pace with consumer spending (i.e. gone up by say 50%) then the volume of computers sold must now be about 10 times higher than in 2000. Which perfectly fits the intuitive pattern you describe, and means elasticity must be significantly greater than 1.
One might doubt whether the increase is really as much as tenfold; if it’s not, then you are right that the hedonic adjustment might be a little too big. But surely that argument is the reverse of the one you’re making?
Wednesday ~ June 15th, 2011 at 9:45 am
Anon
Pardon my “stupid question,” but how do these scenarios differ? In the second, the engine mfrs. buy the software, and better engines for all. Knowledge still spreads, too. When you figure out a better inspection algorithm and code it, other folks will reverse engineer it (if protected by copyright) or read all about it (if protected by patent). They can then incorporate their own improvements or perhaps leapfrog your idea, same as before.
Also, its pretty rare that one person figures out and codes something like that. One of your coders gets poached, and knowledge spreads. (whether or not there’s a trade secret issue there is the same as in your first scenario). Poached coders or poached inspectors…the same knowledge spreads through the human channel, too.
It seems to me (perhaps in my on naivete) that, if anything, knowledge spreads faster in the second scenario. Downloading a new app for an inspection machine takes seconds.
Friday ~ June 17th, 2011 at 12:32 am
Nick Bradley
To answer the title of your post, yes…IP does ‘eat’ TFP:
In your classic growth model, a process improvement spills over to all participants in the market. The cost of adopting the new technology is limited to the cost to train up your employees and purchase the equipment.
With IP, gains from labor/energy/capital/time-saving devices are captured by the owner of the IP license.
– With a higher price, technology adoption slows.
– Since it’s often difficult to build on another person’s patent, innovation slows.
– Through trade secret enforcement by the states (uniform trade secrets act), innovations by firms stay within firms until it slowly leaks out over time through employee turnover. This keeps prices up and margins higher.
Without Intellectual property, TFP growth would be a lot higher and we would probably see deflation most years. Once a new idea is discovered, EVERYBODY would know about it and adopt it…and improve on it.