The AI-powered Veblen economy

There has been much debate online recently about post-AGI economics. For example, Philip Trammell and Dwarkesh Patel made the case for taxing capital under the assumption that once AIs can do jobs better and more cheaply than humans can, the latter will be immiserated unless we do large-scale redistribution. The argument goes something like:

• Baumol effect arguments say that as long as there is some human labor needed in this rapidly growing AI economy, demand for that labor will keep wages high.
• But eventually, enough automation will happen that labor will no longer be a bottleneck, so wages will crash. In fact, as robot labor efficiency goes up, total demand for it might even increase (an idea known as Jevons Paradox).

But the above argument presumes that the eventual consumers of goods don't care whether those goods were produced by humans or machines. In what follows, I argue that we have good evidence that's just not true.

In the mists of time...

In the modern economy, wealth is defined by money. Generally this means ownership of assets which have a well-defined price. But deep in human history, before the Industrial Revolution, before the rise of a burgher middle class in Europe, this wasn't the case. For one thing, economies were small, local, and inefficient. There was a chronic lack of money (in the sense of productive capital). And also, there weren't many real assets like buildings, and those assets that did exist didn't have much of a market for them. A duke couldn't really sell his castle, tied as it was to his title and his power to tax and control peasants. Money just wasn't a thing in pre-modern economies.

So how did they measure wealth? Well, for millennia in almost all human societies, wealth was largely defined by whether you were a servant or whether you were a master. Almost definitionally, poor societies don't have a middle class. Even the term "middle class" suggests an odd, new category between the rulers and the ruled. For most of human history, the middle class didn't exist, so you were in one category or the other, your fate largely decided at birth. Wealth was measured quite simply by how many people you bossed around. Of course, the easiest people to boss around are kids, so this partly explains why people in poor societies tend to have lots of kids if they can. Kids were a cheap source of wealth.

This simple arrangement partly explains why money wasn't so important for organizing ancient and medieval economies: when you know who your boss is, and when you know who your underlings are, there's little need for a general means of exchange.

The modern economy

Our economy works very differently. Instead of having a hard hierarchy of power over each other, we operate more like a network. We exchange goods and services with each other. I pay my tax preparer to pore over the arcana of the US tax code so that I don't have to, and in turn he indirectly pays me to use the software that I write. This economic network allows for specialization, among many other things, and empirically it leads to richer more productive societies. Of course, this comes at the cost of complexity, which is why we had to invent money (and finance in general) to make it all work.

In a very real sense, we're all wealthier than even the richest ancient was. We live longer and we have more material ease: no more cold showers, no more blindness when cataracts start to cloud the lenses in our eyes, no more dying in childbirth, or of a random ear infection. All of those simple things that we take for granted were completely inaccessible to even the richest person in the Middle Ages. And yet, in the unforgettable words of Louis CK: "Everything is amazing... and nobody is happy."

To me it seems clear that a big reason Louis CK is right is because humans crave status.

"Being wealthy beyond the dreams of a medieval king's avarice is great and all, but who am I better than?"

In more mathematical terms, we seem to care a lot about ordinal ranking, so much so that rich societies with little material scarcity need to create it in order to play status games. And as we've gotten richer, the status competitions have gotten fiercer. Some have argued that this is necessary in order to properly compete with each other for the one resource that's still scarce: quality mates. The animal kingdom has no end of zero-sum status competition for mates, the argument goes, and we're no different. No doubt that's part of an explanation, but it doesn' feel load-bearing enough.

One of the most interesting, perplexing, and enduring facts about societies that get rich is that their fertility drops. You may think this is a recent phenomenon, but it's been going on for at least a century. The trend is clear in the total fertility rate (TFR) data. So if becoming rich caused a society to have more resources, time and ability to search for mates, then why don't we live in a Malthusian world where rich people have zillions of kids?

I don't claim to have an answer other than to note that:

• This is a strong empirical regularity in the world.
• Having kids is less attractive if wealth is less about bossing people around and more about usable capital.

At any rate, it seems status competition isn't primarily about finding, attracting and securing mates. In fact, at least anecdotally, the places with the lowest TFR (large cities) are the sites of the very strongest status competition. So, both locally and globally, status competition isn't about mating. Again anecdotally, the less a status competition is about securing a mate, the weirder and esoteric it seems to become. Jimmy Donaldson (MrBeast) has gotten incredibly rich and famous merely by pitting people in weird zero-sum competitions and showing the results to the world. Many of the status games we play today would be unrecognizable to a Victorian, but they're status games nonetheless.

A future AI-driven economy

Perhaps you can see where this is going. We may indeed be heading to a world where AIs and robots dominate humans in economic output (i.e. doing every job better than every human, at a cheaper cost). But I suspect that still doesn't spell the end of gainful human employment. Consider the Birkin handbag.

In a chance 1984 meeting between the CEO of French luxury goods maker Hermès and the actress Jane Birkin, the latter complained about her inability to find a leather handbag that met her specifications. Hermès decided to produce a bag for her, and over time it became an incredibly powerful status symbol. For one, not just anyone can buy a Birkin. One has to be a long-time buyer of other Hermès products before the company considers you a worthy enough client to buy a one. Opaque waiting lists were measured in years. There are also many variations on the bag, some more exclusive than others. With prices ranging from $10,000 to half a million dollars, the Birkin bag is a classic example of Veblen good.

Thorstein Veblen was an economist and sociologist who worked in the early 20th century. Among his many achievements, he coined the term conspicuous consumption, that is, consumption whose primary intent is to signal status. A Veblen good is one that inverts the traditional demand curve, wherein a higher price for the good increases demand for it. One of the key characteristics of such Veblen goods is that they are positional, or ordinal. What matters is what the good signifies about the social status and rank of its owner. Its value derives precisely from scarcity, whether natural or manufactured.

Scarcity is the point of a Birkin handbag. If there were a way to make them automatically then supply would increase and prices would drop. Since it's a Veblen good, total profits would go down. Claims like "It took 250 skilled craftsmen hours to make" are necessary for the bag's status signal to work. Exclusivity and the man-hours it took to make are the source of its perceived value.

As far as I can see, the disagreements over whether AIs and robots will immiserate humans by taking all our jobs really just boils down to this question: will humans be able to culturally create enough scarcity, enough Veblen goods, in order to employ enough people to produce them?

Wither the craftsman?

Of course predictions are difficult, especially about the future, but my intuition is that the answer is yes, we're still going to want human-sourced goods and services. So far, we humans have shown an incredibly creativity in manufacturing scarcity for the purpose of playing status games. If anything, this ability is increasing and the trend is accelerating. Our cratering TFR is only going to enhance this trend, as humans become progressively scarcer. If wealth and status again become defined by who we can boss around, it'll be hard to show how's boss when there aren't as many people around. Scarcity creates value. We might even start seeing upticks in TFR!

Once we're all richer than even the dreams of modern avarice, I suspect we'll still be thinking like our ancestors did. We'll be surrounded by material abundance, but what we'll cherish is that $150,000 coffee cup we bought on our trip to the planetary colonies, painstakingly made by hand from Martian clay by another incredibly rich human.