Karl Smith at Modeled Behavior has a great reply to my post on the GM bailout that features “non-zero orthogonal information,” probability measures, and a hypothetical telepathic dog.
I think the essence of his first point is that no matter how strong one’s overall beliefs about government intervention in the market, that the results of the GM bailout still provide some information that should contribute to how he should see the world. I agree.
I further basically agree with Smith that “the real problem is that the information about GM qua GM is so low that there is a good chance that it is swamped by this bias.” The way I would put this is that we know only the state of the world as it actually exists in the presence of the GM bailout, but the “information” that I really care about is the causal attribution of effects to the bailout. Knowing these causal effects would require us to estimate what the counterfactual world without a GM bailout would look like. My argument is that since we are so poor at estimating this counterfactual world, therefore (to use Smith’s terms), the information is swamped by the bias. Or more precisely, that we are incapable of conducting analysis that should convince rational people who start with different biases to come to a common view of the effects of the bailout.
To be practical about this, Paul Krugman believes that “the auto industry…probably would have imploded if President Obama hadn’t stepped in to rescue General Motors and Chrysler.” I disagree. Until we agree even roughly about this counterfactual world, we can’t agree about the effects of the bailout. But there is no method of analysis that we both accept that can be used to even roughly estimate this counterfactual world, so we’re stuck just disagreeing.
Smith’s second point is that this recognition about of our ignorance calls for “dovishness”:
That is, it calls for being reluctant to accept near term harm for long term benefit. Things that are close up are easier to see. Entropy expands with the arrow of time.
This mediates in favor of being less hawkish on war, less hawkish on the deficit, less hawkish on climate, less hawkish on campaign finance reform, less hawkish on health care, etc.
I also agree with the basic thrust of this, though I would put it as “humility,” or in practical terms, hedging our bets whenever possible. And further, I think it is important to recognize that this applies only at the maximum level of the political hierarchy with which we identify. That is, I think the American government should hedge its bets whenever possible. But trial-and-error improvement within the American political economy calls for sub-entities (say, sates or individual companies) to commit to specific positions, sometimes without hedging.
His third and final point is that while recognizing our ignorance, we need to remember that there is no such thing as “no policy,” saying by example:
One cannot have no tax policy. Even a policy of zero taxes is a tax policy. Even the policy of zero change in taxes is a tax policy.
This is true, of course. What I think recognition of our ignorance leads to, however, is the resulting recognition that what economists and others often call “status quo bias” should more appropriately be called “rational status quo preference.”
If we believe that the current state of a society represents, in part, the current state of an evolutionary process in which functional forms will tend to survive, and further, that various parts of the social organization interact in ways that we do not understand, then both of these observations should lead us to be open-mindedly skeptical of change. This doesn’t mean that all change is bad. In fact, as long we believe that social evolution is eternal, we should accept that any attempt to maintain stasis would be deadly to the society. Some change is essential. But acceptance of our ignorance calls for the burden of proof to be placed on those who advocate any specific change.
To take Smith’s example of tax rates, there is something special about the current tax rate as opposed to all others — it is one part of an organic society that has survived so far, and we don’t really understand what it is about the society that creates this success. Obviously, it’s never really this simple — for example, is the relevant “current state” of society today’s tax rate; or is the specific procedure by which we establish tax rates, which could lead to any given rate; or is the process by which we establish procedures for setting tax rates, and so on up the ladder of abstraction? But at the level of generality of Smith’s reply, this is a rough principle which I think derives from a stance of epistemic humility.