The Lucas Critique is the idea that when policy targets a variable that is supposed to be correlated with a desired outcome, that variable will become decoupled from the desired outcome. That is because by targeting the variable the rules of the game have changed, and with new incentives the variable may no longer be tied to the same causal chains. This critique was originally applied to large scale macroeconomic models that relied on extrapolating observed relationships after one variable is manipulated via policy, but the critique applies more generally as well.
I've posted before about today's disconnect between innovation and patents, but I didn't realize how directly the Lucas Critique applies. Policy was actually set up to increase patents in an attempt to revitalize the economy. Gordon Crovitz explains in his WSJ column: (Hat tip: MR)
Today's patent mess can be traced to a miscalculation by Jimmy Carter, who thought granting more patents would help overcome economic stagnation. In 1979, his Domestic Policy Review on Industrial Innovation proposed a new Federal Circuit Court of Appeals, which Congress created in 1982. Its first judge explained: "The court was formed for one need, to recover the value of the patent system as an incentive to industry."
At the time, policy advisors believed that patents were correlated to economic productivity. This set the stage for today's world, where the correlation between patents and productivity might now be negative. That's because so many of today's patents end up functioning as primarily as a tax on the companies who are moving into new markets. Google spent billions of dollars buying Motorola for its patents to protect its Android ecosystem so it would be armed in the patent wars. Money is being spent on patent portfolios, patents which often describe absolutely trivial ideas, when this money could otherwise be spent on real research (or at least buying companies that did real research and have a functioning product to show for it).
From 2007 to 2011, the CBO found that 89% of the increase in patent lawsuits come from software patents. A recent bill by the House of Representatives targets patent trolls (20% of the lawsuits come from non-practicing entities), but after IBM and Microsoft lobbied to protect software patents the bill does little to stop the patent arms race between large corporations.
The Supreme Court is hearing CLS Bank v. Alice Corp soon, a case about whether a computerized escrow process is patentable. Reform seems very unlikely to happen congress right now, but it will be interesting to see what comes out of the Supreme Court.