China doesn't sell below marginal cost. They sell below average cost, ie cost including recoupment of fixed costs. They've structured their economy that way, because externalities of production mean its best to treat fixed costs as partially a public good. drafts.interfluidity.com/2024/08/13/c... 1/

We recognize that to a degree. We don't think it's weird that states will pay to build a new road, in order to support a new factory and lots of jobs. Why then is it weird if they pay in part for building the factory? What fundamentally is the difference? /fin

In a healthy economy, business profits do not perpetually rise. Individual business profits rise on innovation, then fall to merely the opportunity cost of shareholding entrepreneur time and capital goods. 1/

Whether the profits are paid out or retained doesn't matter very much, investors take their return as price appreciation (earnings retained or buybacks) or dividends. (It might matter importantly to overall economic efficiency, so long-term, but not in an immediate accounting sense.) 2/

Business value in aggregate grow perpetually, but the question is what rate of growth is reasonable to expect (and what kind of growth is reasonable for slowly actively managed baskets of firms like the S&P500, which miss startup levels of growth). 3/

If valuations of S&P 500 firms are held constant relative to profits (let treat them as certain), then either the basket should grow at no more than the rate of GDP growth, or else profits as a share of GDP must continuously grow. 4/

If the profits/GDP continuously grow, it's a constant redistribution of aggregate production from nonshareholder claimants (most notably workers), which seems undesirable. 5/

The only way out of this is to let valuations continuously grow. But that messes up capital allocation. Valuations don't grow for "ordinary firms"—neither VC-ish startups nor S&P 500 stalwarts. "Ordinary firm" profits get discounted at old fashioned rates, interest rates, plus a spread for risk. 6/

Absent some special sauce, expectations of support built into "blessed" firms like those in the S&P, 100x-for-winners expectations in VC land, people evaluate businesses by whether they overcome opportunity costs and penalize them for risk, just like you learned in any finance class. 7/

It is possible — indeed it is current practice — for the state to support continuing valuation growth among a blessed basket of firms. But that creates oligarchs of those in the blessed basket. High valuations means low financing costs. 8/

(No, firms don't often do explicit secondary share offerings. But they issue shares e.g. as compensation, and the higher the valuation the smaller the cost to incumbent shareholders. General debt finance is easier when equity is highly valued.) 9/

So, we can keep an escalator of S&P 500 growth at 10% while GDP growth is 2-3% as long as we want, at cost of creating a two-tier economy, distorting activity towards incumbents + VC-backed firms, incentivizing small firms to join bigger, more highly valued, less competitive agglomerations. /fin

Because tech has proven so oligarchical, so susceptible to absolute domination. Meta doesn't dispute the centrality of AI, so they fear a clear winner will eat their lunch at everything else. (You can see from their a bit silly experiments they think AI has important social media applications.)

It's a really good thing, though. The only I-gotta-hand-it I have for Meta. Linus Torvalds famously described the point of Linux not as "creating financial value" but sucking that value out of operating systems. If he hadn't, we'd all be paying exorbitant taxes to Microsoft and some commercial UNIX.