Negative taxes

 All the last few chats together I would predict we will end up with a system of negative taxes for most people and revenue from corporate taxes.

Yes. And taken together with what we've been discussing, I think that's a surprisingly plausible endpoint.

The basic logic is that if automation becomes extraordinarily productive while requiring relatively little human labor, trying to preserve employment as the principal mechanism for distributing income becomes increasingly awkward. You could instead separate production from income distribution.

A negative income tax is particularly elegant because it doesn't require the government to decide what people ought to do with their lives. Below some income level, the tax system simply becomes a payment system; above it, it becomes an ordinary tax system. In effect, the same apparatus handles both.

Then corporate taxation becomes much more important because the corporations would be collecting an increasing share of the economic surplus created by machines, software, robots and AI. Rather than taxing individual labor heavily—which would be increasingly counterproductive—you tax the capital and profits generating that surplus.

There is an interesting symmetry here:

People receive a basic income through the tax system; corporations pay much of the tax that finances it.

And unlike a conventional welfare state, a negative income tax could be remarkably simple. You wouldn't necessarily need a giant collection of separate programs. Milton Friedman was onto something when he proposed the negative income tax for essentially this reason: it could replace a good deal of bureaucratic means-testing with one continuous income schedule.

The really difficult question, though, is how much of the corporate surplus can be taxed before corporations change their behavior—move capital, retain earnings, reorganize, automate elsewhere, or simply reduce investment. That's where your earlier point about automation becomes crucial. If the productive capacity is sufficiently mobile and international, you can't assume that the government has an unlimited claim on the machine-generated surplus.

But I can imagine a world in which the old division—workers earn wages, investors earn profits, government taxes both—gradually becomes less representative of how the economy actually works. A system in which most citizens receive some net payment from the tax system while a relatively small number of highly productive enterprises provide most of the tax revenue could eventually make more economic sense.

It would be a rather extraordinary reversal of the twentieth-century welfare state: instead of taxing a mass of workers to support a minority who aren't working, you'd be taxing a highly automated productive economy to support a population for whom there simply aren't enough economically necessary jobs.

All taxes have effects, often perverse effects, on behavior.  One goes into such things with lots of study and simulation.

Exactly. The interesting question isn't whether a tax has behavioral effects—it inevitably does—but whether the effects are better or worse than those produced by the alternatives.

And with a future negative-income-tax/corporate-tax system, the modeling would become especially important. You'd want to simulate at least the responses of workers, corporations, investors, and governments. A corporate tax might encourage debt rather than equity, accelerated depreciation, profit shifting, relocation, or vertical integration. A negative income tax might affect labor-force participation, hours worked, marriage decisions, saving, and the willingness to take unpleasant or marginally worthwhile jobs.

There is also a particularly tricky interaction between the two. Suppose automation makes a company enormously profitable and the government taxes that profit to finance the negative income tax. If the resulting income floor allows people to refuse lousy jobs, employers may have to raise wages or automate those jobs. That could actually accelerate the transition toward automation.

But the same mechanism could go too far. If the guaranteed income is high enough, the supply of labor for some socially useful but unpleasant occupations could collapse. You'd then get either very high wages for those jobs or pressure to automate them—and perhaps both.

So I think your phrase "lots of study and simulation" is exactly right. This is one of those areas where political intuition is particularly unreliable. A policy that sounds wonderfully efficient in isolation can produce a second- or third-order response that largely defeats it.

It also makes me think your insurance experience gives you a useful way of looking at this—not because taxation is insurance, but because you're accustomed to thinking in terms of behavior changing when the rules change, rather than treating the population as a static collection of numbers. The model has to include the reactions of the people being modeled.

I suggest you are thinking in overly static terms; changes are introduced gradually, indexed, etc.  The transition would ultimately to be to work only when done for the work and not for the income, but that's generations away.

Yes, that's a fair correction. I was treating the system too much as though someone flipped a switch from today's tax system to the eventual one.

A realistic transition would be much more evolutionary. Benefits and tax thresholds could be indexed, rates adjusted as experience accumulated, and the system expanded only as automation actually displaced economically necessary labor. You'd learn from the behavioral responses along the way rather than pretending that a model produced in 2026 could specify the optimum system for 2055.

And I think your distinction about working for the work rather than for the income is the really profound part. That's quite different from simply paying people not to work.

If material necessities become sufficiently inexpensive because automated production is doing most of the economically necessary work, employment could gradually cease being the principal justification for distributing purchasing power. People might still work because they like teaching, building things, caring for others, researching, making music, running restaurants, gardening, writing, or simply having something useful to do. But the connection between "I must do this or I cannot support myself" and "I do this because I want to" could weaken enormously.

And yes, generations away is an important qualification. Even very rapid technological change doesn't instantly replace the accumulated institutions, expectations, skills, property arrangements and social conventions built around employment. My guess is that the transition would be sufficiently gradual that each generation would regard the next step as a modest adjustment, while historians later would look back and say, that's when the nature of work fundamentally changed.

There's an interesting irony in that: the eventual destination might be quite radical, while getting there could be almost boringly incremental.