Research suggests that both knowledge and thinking style contribute to better forecasts, and likely to more accurate empirical beliefs more generally.* Lately I’ve been thinking about how microeconomics might help explain how those two inputs fit together.
(Disclosure 1: I’m sure papers have been written about this and I’d love to hear about them! Disclosure 2: I’m not an economist. Just someone who’s taken microeconomics and thinks too much about the research on belief accuracy. Apologies if I got something wrong!)
In an introductory microeconomics course you learn about production functions, where the amount a firm produces varies as a function of just two inputs: labor and capital. Such a function could in theory take many shapes, but here’s one example:
q(L,K) = √(LK)
What if we think about a production function for accurate forecasts or understanding of the empirical world more generally? We could think about understanding of the world as a function not of labor and capital but of knowledge and thinking style (wisdom), per the research by Philip Tetlock mentioned above. In this model, “understanding” isn’t a classic belief-accuracy measure, but a level of production. You’re choosing how many “units of understanding” you want to produce. (I’m imagining an individual as the producer, even though this model is typically used to represent firms.) You can always choose to invest more and get some marginal additional return to your understanding. Here’s how such a function might look:
U(K,W) = √(KW)
(Yes, my letters here are terrible, overlapping as they do with commonly used ones in economics.)
- U = Understanding (not utility)
- K = Knowledge
- W = Wisdom (thinking style)
What does this model suggest that might be interesting when pondering the nature of rationality and accurate beliefs?
First, in these models the “firm” sets marginal revenue equal to marginal cost, and in a competitive market marginal revenue is the same as price Here, the individual keeps producing understanding until the marginal cost of doing so equals the price someone is willing to pay for that understanding. More intuitively, the higher the payoff to understanding something, the more you’ll invest in understanding it. The more expensive the inputs — acquiring more knowledge and a better thinking style (wisdom) — the less you’ll invest.
Second, I picked the formula above because it has both overall diminishing returns — every unit is less efficient to produce than the last one — and diminishing relative returns between the two inputs. You’d rather have an equal mix of knowledge and wisdom than lots of knowledge and no wisdom, or vice versa. I don’t know of any research suggesting how these two variables interact, but the idea of balancing them rather than going “all in” on one or the other seems interesting.
Third, the assumption when you learn this model is that capital is fixed in the short term, meaning that in the short term firms only decide how much labor to employ. Only over the long-run are they free to invest in more capital, and the freedom to mix the two usually results in greater efficiency.**
This set up seems to have a parallel in terms of understanding and belief accuracy. For an individual, thinking style (wisdom) is fixed in the short run, but not in the long run. If you want to understand a topic more by tomorrow, you really only have one choice: try to learn more about it. But if you want to have a greater understanding by next year, you might invest in improvements to your thinking style — your “mindware” in Keith Stanovich’s terminology — by learning a bit of probability or taking “calibration” quizzes to calibrate your confidence in your beliefs to your accuracy.
This is all speculative, of course. Just because you can write down a model doesn’t mean it’s how things actually work. The reasons people acquire understanding (or don’t) can’t be explained solely by markets and economic self-interest. But thinking about forecasting and beliefs in the context of microeconomics at least yields some interesting ideas and hypotheses.
*Other things, like intelligence, help, too. But I’m leaving them out for simplicity.
**Depending on the shape of the production function and input prices.