I come to bury industrial policy, not to praise it (pt I—Lerner)

A recent  renewal of public discussion regarding the virtues of industrial policy, juxtaposed against the ongoing Solyndra spectacle, prompted me the other day to revisit an Economist debate from last year between Josh Lerner and Dani Rodrik. The debate was on the following motion:

"This house believes that industrial policy always fails." 

Lerner, whose expertise is mostly in rich country (US & Europe) innovation policy, argued for the motion; Rodrik, a development economist, argued against.

This Lerner-Rodrik debate is worth revisiting for two reasons. First, Lerner and Rodrik are exceptional intellects with imposing publication records, debating an important topic. Second—most entertainingly for me as the guy in the bleacher seats—they both manage to be wrong, in spite of having ostensibly staked out opposing positions.

Let's start with Lerner.
Here, as when he first wrote on this topic over a decade ago, Lerner focuses on what he characterizes as "the disastrous history of most loan programmes to finance high-growth entrepreneurial businesses." To evidence the disasters, he points to two problems:

1. bureaucratic incompetence as evidenced by failure of programs to be self-sustaining, because they demand too little of the few "winners" and thus are rapidly depleted as a consequence of the majority's losses, and
2. "regulatory capture" whereby "programmes geared towards boosting nascent entrepreneurs may instead end up boosting cronies of the nation's rulers or legislators." 

Lerner eloquently articulated his perspective with regard to bureaucratic incompetence (part 1 of his argument) in a 2001 paper titled "When Bureaucrats Meet Entrepreneurs: The Design of Effective 'Public Venture Capital' Programs" (p. 81):

The initial reaction of a financial economist to the argument that the government needs to invest in growth firms is likely to be skepticism. A lengthy literature has highlighted the role of financial intermediaries in alleviating moral hazard and information asymmetries. Young high-technology firms are often characterized by considerable uncertainty and informational asymmetries, which permit opportunistic behavior by entrepreneurs. Why one would want to encourage public officials instead of specialized financial intermediaries (venture capital organizations) as a source of capital in this setting is not immediately obvious.

Now no economists is going to dispute either that financial intermediaries play a role in "alleviating moral hazard and information asymmetries," or that "young high-technology firms are often characterized by considerable uncertainty and informational asymmetries." No disagreements there. But there are real problems with the following sentence: "Why one would want to encourage public officials instead of specialized financial intermediaries (venture capital organizations) as a source of capital in this setting is not immediately obvious."

First of all, venture capitalists have long conceded that they're not particularly interested in seed stage, technology-based ventures. (See e.g. this 2001 piece [p. 104] by venture capital legend David Morgenthaler, or just Google the topic to sample the veritable avalanche of substantiation that has been published since.) As the median size of venture capital deals grew beginning in the mid-1990s and pressure on managing partners to provide attractive returns to investors in mammoth funds intensified accordingly, venture capital tended increasingly to flow to projects in later stages of development, to already-proven technologies, or to other investments with potential be rapidly "flipped." But a real science-based company—one of the few that aspires to accomplish the difficult transition from invention to innovation—can take a decade or more to get to the point where it earns revenue, much less a profit. Given such technical and market uncertainties, venture capitalists and even many individual "angel" investors understandably preferred to avoid funding speculation. The very success of the venture capital industry has driven it away from funding technology entrepreneurs in the early stages. High "hurdle rates" for expected returns to venture capital investments translate into high hurdle rates for reading business plans. The gap between the resources required by early stage technology entrepreneurs (typically, in the $100K to $2 million range) are inconsistent with the deal size required by mature venture capital firms to support operations given their current scale of operations.

Here's the more fundamental point: Government investments in potentially high-growth firms is good idea precisely because of the fact that Lerner stresses: the "specialized" skills of financial intermediaries such as venture capitalists are not easily imitated, and thus are in short supply. Why? Again, as Lerner notes, early stage development involves not only high quantifiable risks, but also daunting uncertainties. When the uncertainties are primarily technical (as is the case in science-based innovation), garden variety bankers and fund managers are ill-equipped to quantify them. For new technologies that have the potential to create new product categories, market uncertainties are also high and similarly difficult to quantify. The "due diligence" that investors in venture capital funds require of managing partners and that angel investors require of themselves is intrinsically difficult—and getting more so as both technologies and markets become increasingly complex.

As Lerner himself implies in the above quote, simple labor economics suggests that a public official on a government salary will be less capable of evaluating the market prospects of a new technology than will a highly paid venture capitalist. To suggest otherwise is, rightly, to invite derision. However, the same set of basic principles suggests that venture capitalists are, typically, highly compensated because their skill set is not easily imitated. Barriers to entry and high compensation both imply that the venture capitalist's opportunity cost of time is much higher than that of the public official. For that reason, the efforts of public officials to evaluate the commercial dimensions of technical proposals—for example, the review of applications for competitive government awards such as those given by the U.S. Small Business Innovation Research (SBIR) program—are better understood as complements to, rather than substitutes for, the due diligence performed by venture capitalists.

Differences between venture capital and public modes of support for technology entrepreneurs go beyond both the capabilities of specialized intermediaries as compared with those of public officials and the quantity of funding sought. As work by David Hsu (and others since) has documented, entrepreneurs who seek venture capital investments are typically seeking more than funding. They also seek mentoring, access to contacts, and strategic guidance. In return they relinquish a degree (sometimes large) of control over their organizations. In contrast, a technology entrepreneur seeking a government award can not expect to benefit from mentoring and other such forms of support. At the same time, however s/he can be assured that his/her equity stake will not be diluted. The risk/reward profile of the two forms of financing is thus sufficiently different to call into question the appropriateness of terms such as "public venture capital" frequently used to describe various modes of public financial support to technology entrepreneurs. While the funds supplied may be public, the mode of support bears very little resemblance to venture capital.

To the extent that the sort of there are systematic and persistent impediments to the funding of potentially high-growth, technology based firms, they do so—again!—because the skill set of individuals specialized in evaluating linked technical and commercial opportunities is very difficult to imitate. For this reason, varied contractual arrangements coexist, each offered within one of a diverse set of institutional contexts. Taken together, in a vibrant technology ecosystem, these arrangements provide technology entrepreneurs with a menu of options each with a distinct risk/reward profile. Design and implementation of effective policies to support technology entrepreneurs begins not by looking at one or another category of support in isolation, but rather with an understanding of, and appreciation for, the entirety of this menu of options.

As for the "regulatory capture," the debacle surrounding the Solyndra bankruptcy would seem to provide definitive evidence that programs to support high-growth companies are just as subject to political gaming and cronyism as any other large-scale government energy initiative—be it a policy to support corn-based ethanol or (if you want to see some real money) one to give oil companies breaks on off-shore drilling leases in the name of "energy independence." And, that reality is, the answer is that the Solydra debacle does provide such evidence ... at least, for a program handing out loan guarantees worth half-a-billion dollars to a single company. However, such a program simply cannot be compared with the Small Business Innovation Research Program, which awards competitively administered grants (not loans) that average one thousandth the size of the loan guarantees for Solyndra. The SBIR is relevant because it is the program that Lerner actually studied, originally, in deriving the lines of argument he advanced in The Economist debate. 

As it turns out, I'm also quite familiar with this program, having spent three years as part of the research team for a National Academy of Sciences study of its effectiveness. (I was the research lead for the Department of Energy report). In the actual context of the SBIR program, evidence of "regulatory capture" usually focuses on a few "SBIR mills" who receive a disproportionate share of grants without demonstrating real commercial impact--ostensibly proving that they're gaming the system. I studied one such company while I was working on the of the National Academies study--Creare, based in Hanover, New Hampshire.When visited Creare I saw lots of geeks with PhDs, and not too many crony-looking types. I ultimately learned that Creare had incubated more than a dozen spin-off companies, with aggregate sales at that time of over $300 million. None of that impact would have shown up as "commercialization" in the terms employed by Lerner in his original, and still most serious work on this topic.

So the problem with Lerner's argument is that he happens to direct his attention just the one type of  "industrial policy" that is actually most likely to work: competitive, professionally evaluated awards extended to early stage companies in countries with highly developed private equity markets. Can such programs be worth while? Yes. If they count as "industrial policy," then the motion fails.

Now as for Rodrik's argument against the motion...

This post draws from Philip Auerswald  (2007) "The Simple Economics of Technology Entrepreneurship: Market Failure Reconsidered," in David B. Audretsch, Isabel Grilo and Roy Thurik eds., The Handbook of Entrepreneurship Policy, Northampton, MA: Edward Elgar.