It's no secret American manufacturing is in crisis, and that its problems form a significant component of our present economic mess. I've written before about how the Obama administration may (may!) be starting to get serious about the problem.
Another small but significant data point on the question of whether the administration is serious took place this last week: the government's new National Network for Manufacturing Innovation held its first conference, designed to elicit public input on how this program will be designed and run.
The event was held at Rensselaer Polytechnic Institute, a good engineering school in Troy, NY, and while I was not there myself, the head of my organization was, and he debriefed me on what went on. (The NNMI website is here. RPI's page on the conference is here. Click here for a pdf of the conference schedule.)
Now the details of NNMI haven't yet been settled, so I can't comment on them. But it is possible to know, even at this early juncture, that if it is to succeed, its policies must rest upon understanding and implementing a correct vision of its economic rationale. Absent this, it is likely to either fail to generate economic benefits to the nation, or collapse outright in a flurry of Solyndra-style scandals. There is a huge temptation to just declare manufacturing "holy," as environmental technology was previously declared holy, and throw public money at it.
So what should NNMI do instead?
The first rule of industrial policy is that "create jobs" is not a valid strategy, despite the political appeal of this concept. Anyone can spend $1.00 to create $.75 worth of jobs. The problem is that the $1.00 has to come from somewhere--it has to be taxed, borrowed, or cut from other spending. And $1.00 in the public's pockets will, other things being equal, create $1.00 worth of jobs, not $.75. As a result, while benefits may be apparent, they will not be real.
Effective industrial policy depends upon finding uses for $1.00 that will somehow create more jobs than would have been created if the money had just been left with the public. This can be done: the idea that it is impossible is ultimately identical with the proposition that markets are perfectly efficient: a known falsehood in other areas of economics and ultimately an ideological dogma.
Government has a number of legitimate roles to play here, and many of them are quite complex. But most of what is fundamental boils down to solving two key problems that the private sector cannot solve on its own:
- Appropriability, or the fact that many useful innovations are difficult for the innovator to capture the full economic value of.
- Time horizons, or the fact that the private sector won't invest in projects whose profits, although appropriable, are too far in the future.
So-called infratechnologies fall into the first category. These are technologies, like the Internet, which enable a huge number of profitable innovations but which are themselves, for various reasons, hard to make a direct profit off of. As a result, the free market tends to under-supply them, and there is a strong prima facie case for the government to fund their development.
To take one example, commercial nanotechnology companies depend, according to Greg Tassey of the National Institute of Standards and Technologies, upon the following key infratechnologies:
- Techniques for measuring the shapes, dimensions, and electrical characteristics of the various molecules making up nanoscale devices.
- Techniques for manipulating and measuring the spin of individual electrons.
- Scientific and engineering data for characterizing the fundamental physical behavior and long-term reliability of new nanoelectronic materials.
Mainstream neoclassical economics assumes (often tacitly and without even realizing the issue exists) that new technologies grow automatically from advances in pure science. It also assumes that new technologies automatically commercialize themselves. But both these assumptions are observably untrue, largely due to appropriability and time-horizons problems.
Historically, the U.S. solved the problems of appropriability and time horizons by indirect means. We privileged certain oligopolistic sectors of corporate America to reap exceptionally high profits in exchange for developing technologies that would otherwise probably not have been developed.
Some of this was done by way of defense contractors, some by way of very large companies with monopoly or quasi-monopoly power over their ultimate product markets. Thus the old AT&T with its Bell Labs, the old IBM with its Watson Laboratory, the old RCA with its Sarnoff Research Center, the old Xerox with its Palo Alto Research Center, or GM in its glory days.
Because of these companies' oligopolistic power, they were assured of a) capturing the value of whatever they discovered or invented, rather than having it swiped by a competitor, and b) bringing in enough money, over a long-enough time frame, to pay for expensive laboratories that could take many years to produce results.
Unfortunately, these companies are largely gone, or so internationalized that they confer no especial benefit upon the U.S. economy, as opposed to any of the other nations where they do business.
Worse, because the U.S. solved the problems of appropriability and time horizons indirectly, there never crystallized an explicit ideological consensus in this country about these being the key rationales for active industrial policy. Indeed, to a huge extent, we fooled ourselves into thinking that our national economic success was caused by our (fictional) embrace of extreme laissez-faire.
Contemporary venture capitalists almost never operate beyond a seven-year time horizon. (Thus we observe that the technology underlying Google was developed from research funded by the National Science Foundation on digital libraries.) For all its very real achievements, the venture capital system is largely a system for harvesting fundamental innovation, not creating it.
It follows that the key question that will need to be asked, whenever NNMI considers funding some project, is whether it is being asked to fund something that the private sector should be funding on its own. (Solyndra clearly fell into this category, as there were no appropriability or time-horizons issues presented in their business model.) Instead, NNMI should seek out projects that have the following characteristics:
- They involve developing technologies where much of the benefit will "leak" to parties not compelled, by patent or other regulation, to help defray the cost of developing them.
- They involve developing technologies whose payoff, though substantial, will occur beyond an approximately seven-year time horizon.
These two key issues are a highly abstract description of the problems involved, and they ramify enormously and interact with other issues--giving rise, for example, to the notorious "valley of death" problem in innovation. So they should not be misunderstood as exhausting the concerns here. But getting these issues right will be fundamental to any successful active industrial policy.