Goods analogies

Analogies are useful to reason about new things. Goods are good analogies for each other when they are close in some suitable sense. So what makes a notion of closeness “suitable” to reason with?

What should distance measure?

When I want to know whether two products are “similar” I have in mind someone who is buying or selling them. Regulators, investors, public commenters–anyone with an interest will usually want to consider at least one of those perspectives. I like to think of distance metrics over products as reflecting demand side or supply side factors. “Demand side” distances over products reflect the preferences of their purchasers, and “supply side” distances over products reflect the capabilities of their producers.

Products that are closer in demand space are better substitutes or complements for each other. Products that are closer in supply space have more shared production elements. Picture a plane of pairwise distances between products, with “demand side distance between products” on the x axis and “supply side distance between products” on the y axis. Products that are close in demand space but far in supply space (the upper left quadrant), or far in demand space but close in supply space (the lower right quadrant), may present opportunities to sell into or source from new markets.

Is this worth calculating?

Maybe sometimes? I always liked the idea of analogy weighting. It seems like it should be possible and maybe worthwhile to do something similar to measure demand and supply distances between products in a “large market” context like a supermarket or national retail chain. You need many subtle interactions and decisions to justify the machinery for scalable inference and prediction. It’s probably not worth doing formally in “smaller” (most) markets, though the intuition holds. Case studies or specific analogies may be more useful then.