How Much Economic Activity Is Cross-Subsidized?

There's a fun and popular game of looking at a store that seems economically implausible – say, it only sells expensive candles, and not very many of them – and asking yourself, how the hell does that thing stay in business?

I was discussing this with friends recently and was surprised to discover that not everybody shares my answer/assumption: that the shop is losing money, but the store-owner has a wealthy spouse or parent, and the shop is cross-subsidized by the spouse-or-parent's finance/tech/industrial money.

I realized that this assumption has been lodged in my brain for years, but that I don't actually know how true it is. And that I also don't know of any way to test it: I searched online and couldn't find either data or case studies.

For me, part of the background here is a broader question about whether you can infer that something is a "good business" from seeing that other people are doing it and not (seemingly) going bankrupt. And this is very much impeded if lots of businesses you see are actually cross-subsidized by something else: that means you can't learn from their ongoing existence that the business is profitable, they're just willing to lose money that you might not be.

Of course, cross-subsidization can exist within businesses as well as across them, so the same argument applies at fractal scales: just because a business is doing a certain project, doesn't mean the project makes sense, they may just have other projects inside the business cross-subsidizing this one.

Anyway. This blogpost doesn't have a conclusion, it's meant more as a batsignal: if you know of any data about cross-subsidization of businesses, or any ideas about how I could find such data, please send it my way.