Only Theory Can Say What Assumptions Imply - A Reply to Rice

Robin Hanson, Cohort IV, for RWJF SHPRP '98 Meeting, Aspen

As a health economics theorist, I am delighted to see Thomas Rice direct our attention to some interesting alternative assumptions. I not only think these assumptions worth further investigation, I have personally explored many of their implications for health policy. I also have no objection to Rice's reminding a wider audience of GIGO (garbage in, garbage out); conclusions coming out of a theoretical analysis are limited by the quality of the assumptions that go in. And since models are usually oversimplified for clarity, we should be cautious regarding policy inferences based their conclusions.

I am disturbed, however, that

  1. Rice seems to identify all "economic theory" with a specific set of assumptions he dislikes,
  2. Rice's claim that "all important questions must be answered empirically" does not seem to leave a substantial policy role for theoretical models based on better assumptions, and
  3. Rice seems to think that he has offered substantial analyses of the policy implications of his preferred assumptions.
To me, "economic theory" just means applying formal mathematical models to situations of "economic" interest. Economic theorists are open to and have explored a wide range of assumptions, including those Rice favors. Theorists just insist that one show how one's conclusions follow from one's assumptions.

Perhaps health economics theorists have underestimated the relevance of Rice's assumptions. This, however, lends little support to the many policy conclusions Rice makes (see reverse page). Rice offers no formal models based on his favored assumptions, nor does he cite specific models by others. And without explicit relevant models based on these assumptions, it is hard to see how we can say much about their health policy implications. The remainder of this reply illustrates this point by showing how two of Rice's favored assumptions need not imply Rice's favored policy conclusions.

Consumer Ignorance As a response to consumer ignorance, Rice favors having government experts restrict the sale of remedies. I have, however, built a model comparing product bans vs. warning labels. While I do find that even ideally-motivated regulators will ban products from ignorant consumers, I also find a general class of cases where consumers are better off when regulators may not ban products. My proof allows for many forms of altruism, envy, endogenous preferences, and irrationality. Similar models should apply to Rice's cases of banning vs. labeling low-quality health plans, and restricting the supply of health care vs. labeling some types of care as less effective.

Envy/Altruism The policy implications of envy and altruism depend greatly on their specific nature. For example, utility-based envy and altruism greatly weaken the case for in-kind transfers. And appropriate policies depend dramatically on the scope of envy and altruism. If people only care about their co-workers, for example, then individual employers would internalize the problem, leaving little justification for outside intervention. In contrast, a very wide scope in space and time may suggest global wealth redistribution and savings disincentives far beyond what most would advocate.

Also, envy may have evolved from contests to attract mates. If so, status-signaling contests may justify subsidies for health spending invisible to potential mates. Such envy, however, would not apply to those too old or distant to compete for mates, nor to Rice's case of market-driven health services gaps. And if altruistic and "warm-glow" behaviors are primarily signals of character in mating games, even more dramatic policy implications may follow.

The details really do matter, and economic theory shows how.