Monday, October 19, 2009
The technology we don't see
David Henderson points out a passage in a June report on health care reform by the Council of Economic Advisers:
The conversation on this post last week ended up being an argument over this very same question: How do you put values on lives? The point is public policy has to. A statistical value of a life is part of any decision on regulation that goes towards someone's safety or health. To get at that question requires one to observe human choice and infer value on the basis of those choices, assuming those choices are rational. Since there are finite resources and potentially infinite demand for safety some measures simply aren't done. Which ones? How do you choose? Public policy relies on a marginal approach. At the margin, the cost of saving one more life is higher than the benefit received. As Viscusi [1993] posited, there is a 6,000-to-1 chance of someone dying from an asteroid, so why do we not have a "doomsday rock defense" if not for the prohibitive costs of building the gun? Both costs and benefits matter in the decision: There are many things that have a statistical chance of fatality less than 1/6000th, but we do them because the cost of avoiding the hazard is quite low.
The CEA quote I began with misses one other point, though. The first attempts to solve a problem will likely be rather expensive. There is a learning-by-doing aspect at play (in the sense of Stokey [1988]) -- new goods enter and those that provide value at lower cost push the higher cost ones out. But one cannot know what provides value without the entry of some good first, be they higher- or lower-priced. Creating higher cost goods allows one to learn how to produce at lower cost. The CEA seems to support short-circuiting that discovery process. That would be technology we don't see, Bastiat-style.
In medicine, however, technological progress in recent decades has been almost exclusively cost-increasing, without generating a commensurate increase in value. Undoubtedly, provider incentives, which largely reward finding an expensive way of treating a previously untreated condition rather than finding a less costly alternative to an existing treatment, contribute to this trend. (p.20)Henderson writes that surely they don't mean no value. A disease that was untreated previously but now is, must provide value. Perhaps they're not valuable enough using a benefit-cost principle. But how would you know?
The conversation on this post last week ended up being an argument over this very same question: How do you put values on lives? The point is public policy has to. A statistical value of a life is part of any decision on regulation that goes towards someone's safety or health. To get at that question requires one to observe human choice and infer value on the basis of those choices, assuming those choices are rational. Since there are finite resources and potentially infinite demand for safety some measures simply aren't done. Which ones? How do you choose? Public policy relies on a marginal approach. At the margin, the cost of saving one more life is higher than the benefit received. As Viscusi [1993] posited, there is a 6,000-to-1 chance of someone dying from an asteroid, so why do we not have a "doomsday rock defense" if not for the prohibitive costs of building the gun? Both costs and benefits matter in the decision: There are many things that have a statistical chance of fatality less than 1/6000th, but we do them because the cost of avoiding the hazard is quite low.
The CEA quote I began with misses one other point, though. The first attempts to solve a problem will likely be rather expensive. There is a learning-by-doing aspect at play (in the sense of Stokey [1988]) -- new goods enter and those that provide value at lower cost push the higher cost ones out. But one cannot know what provides value without the entry of some good first, be they higher- or lower-priced. Creating higher cost goods allows one to learn how to produce at lower cost. The CEA seems to support short-circuiting that discovery process. That would be technology we don't see, Bastiat-style.
Labels: economics, health care
