Oracle, the CIA and government's role in picking winners

After Larry Ellison's announcement at the end of last week that he was retiring from Oracle, I saw (on Twitter, of course!) a blog post, by someone named Matt Novak, entitled "Larry Ellison's Oracle Started as a CIA Project." Novak noted that official Oracle accounts of the history of the company don't mention that the company's first customer was the CIA, and that in fact the name the agency wanted for the relational database was... Oracle. "Not mentioning just how reliant Oracle has been on government contracts since its inception," Novak writes, "is downright strange and seems to feed this narrative that Ellison simply created a product that companies wanted and private enterprise did the rest."

If you continue to read Novak's post, it becomes apparent that his aim is less to praise the CIA for helping a premier American company get off the ground than it is to tar high-tech companies for supporting government surveillance and hurting people's privacy. But my reaction to the post was somewhat different -- it made me think again about a question I think has gotten insufficient attention from scholars and others, namely the role of government in promoting new industries and long-term economic growth.

It is (relatively) uncontroversial among economists to believe there is a legitimate role for government in funding basic research (research prior to product development, that advances general scientific knowledge), on the argument that private firms cannot recoup all the value that scientific discoveries create. On this view, the huge government funding for medical research helps explain the prominence of our pharmaceutical and biotech industries, much as government support for basic computer sciences and hardware-related research helps explain the booming high-tech industry.

But the Oracle example seems to suggest something additional going on. It has often been noted that three of the most successful American industries -- high tech, aerospace and agriculture -- are ones where the government has not just sponsored basic research but also gotten significantly involved further downstream: buying company products when they were just coming to market, actively supporting their development, or spreading practical knowledge to the industry.

This comes dangerously close to "picking winners" -- something most economists, and certainly those on the more free-market side of the political spectrum, don't like. Most economists would say government tends to do a bad job of that, and this should be left to the marketplace.

I find the argument economists make convincing in theory. I am less convinced, however, that it always holds up in practice. I don't have a fully formed theory, but I think academics should work to develop one that predicts when actions like the CIA supporting Oracle are likely to be good for the economy, and when such efforts will be ineffective or counterproductive.

One view would be that the government should not support a startup as a way of "picking winners," but only when it has a clear need of its own to be filled. As the blog I read noted, the CIA truly needed powerful relationship databases. The Defense Department needed military aircraft independent of the impact of its purchases would have on the development of the aerospace industry. In such cases, government purchases of technology would seem justified, especially since government, like a venture capital firm, is in a better position to take large risks in development even if most risky development efforts fail.

(It's worth noting that, even if every government R&D investment since 1789 had produced zero return except for the Internet, government research money since the beginning of the Republic would almost certainly have been justified.)

Finally, I think contracts to sponsor development or buy developmental products are better than grants or other direct support to companies. I trust the contracting system more to put pressure on a company to produce results than various kinds of direct financial assistance.

These are just a few thoughts. I still think we need a better theory of when government efforts to promote industry are justified and not.