Center for Economic Analysis

For more than two centuries, a single, elegant assumption has served as the circulatory system of macroeconomic theory: supply creates its own demand.

First articulated by Jean-Baptiste Say in 1803, the logic is often explained by stripping the economy down to its barest form, a barter system. In a pure barter economy, you cannot buy something unless you first produce something to trade. If a farmer wants olives, they must produce and supply wheat to exchange for them. Therefore, the very act of bringing wheat (supply) to the market is simultaneously the act of demanding olives. Supply and demand are not just equal; they are two sides of the exact same action.

In a modern monetary economy, this principle translates into a continuous, circular flow of income. To produce goods, a firm must pay the factors of production, paying wages to workers and profits to owners. Those individuals, turn around and spend that income. The act of production inherently injects the exact amount of purchasing power into the economy needed to clear the market.

But what happens to the global economy when the mechanism that sustains this circular flow, human labor, is permanently bypassed? As advanced artificial intelligence and extreme automation decouple production from wage distribution, we are rapidly moving past cyclical economic disruptions into a profound structural paradigm shift.

The Two Great "Wedges" of Economic History

To understand the threat AI poses to our long-standing understanding of the foundation of the macroeconomy, we should look at another historical shift that fundamentally fractured Say’s Law: the introduction of money.

In a pure barter economy, as established, Say’s Law is absolute. The introduction of money split the single act of trading into two separate acts: selling (trading wheat for money) and buying (trading money for olives). This created a "wedge." Because money acts as a store of value, a person can sell their wheat and simply hoard the cash. As John Maynard Keynes demonstrated, this psychological and temporal wedge causes purchasing power to leak out of the circular flow, leading to recessions where goods sit unsold on shelves.

AI introduces a second, far more expansive wedge. It is not psychological; it is structural.

Under automation, the leakage in demand occurs because the mass consumer base never receives the income in the first place. When an automated facility produces goods, the income flows almost entirely to the owners of the capital (the compute, the algorithms, the robotics). Because capital owners have a low marginal propensity to consume (they cannot physically buy enough groceries, cars, and services to absorb the output of an automated global economy) aggregate demand collapses. The engine stalls not from an inability to supply, but from the dismantling of the wage mechanism that allows consumers to demand.

The Ghost of Marx in the Machine

It is difficult to analyze the macroeconomic implications of AI without hearing the echoes of Karl Marx. While modern data-driven policy often eschews classical Marxist theory, the endgame of artificial intelligence maps almost perfectly onto the structural crises he predicted for late-stage capitalism:

• The Organic Composition of Capital: Markets force firms to continuously invest in machinery (constant capital) to outcompete rivals, reducing reliance on human labor (variable capital). The billions currently pouring into AI data centers and model training represent the ultimate realization of this trend, driving variable capital toward zero.
• The Crisis of Underconsumption: Capitalism requires firms to minimize wages for profit yet requires those same workers to be well-paid consumers. By automating labor to maximize efficiency, the system inadvertently destroys its own consumer base, leading to an automated abundance that no one can afford to buy.
• The Permanent Reserve Army of Labor: If algorithms can perform cognitive tasks faster and cheaper than humans, vast swaths of the professional class will be pushed into a structural, permanent oversupply of labor, aggressively depressing wages for any jobs that remain.

Trapped in a Dead End: Why Standard Fixes Fail

Normally, when the economy slows down, central banks step in to fix it. Their primary tool is lowering interest rates to make borrowing cheaper, hoping people will take out loans to buy cars, homes, and start businesses. But an AI-driven economy completely breaks this toolkit.

As wage income evaporates and is transferred to a small group of capital owners, everyday consumer spending plunges. In this scenario, it doesn't matter if the central bank drops interest rates to zero. You cannot coax an unemployed consumer into taking out a loan if they have no underlying wage income to pay it back.

Simultaneously, the extreme efficiency of automated factories will likely make goods incredibly cheap. But a price tag of pennies is still too high for a consumer with zero dollars. The traditional levers used to steer the economy simply detach. The system falls into a trap where central banks become powerless to stimulate demand.

Engineering a New Floor: The UBI Balancing Act

When the traditional ideal, competitive job market that pays a living wage is no longer viable, governments must pivot to sweeping structural interventions.

If central banks are powerless in an automated trap, massive government policy becomes the only lever left to keep the economy running. This naturally points toward a Universal Basic Income (UBI) or a Negative Income Tax. However, designing a functional minimum income is notoriously difficult. Policymakers have to delicately balance three competing "dials":

1. The Starting Floor: The baseline amount of money guaranteed to a citizen simply to survive.
2. The Clawback Rate: The speed at which government support is reduced (or taxed away) if a person does manage to earn outside market income.
3. The Tipping Point: The specific income level where a citizen stops receiving help and transitions into being a net-taxpayer.

Balancing these dials isn't just basic math; it requires massive real-world simulations to predict how humans and markets will react. Interestingly, in a fully automated future, the clawback rate ceases to be about making sure people are still motivated to work, because the work won't be there. Instead, it becomes a pure mechanism for inflation control: heavily taxing the concentrated wealth of automated tech giants to continuously fund the consumer base.

We are standing on the precipice of an economy where wealth is created independently of human toil. Say's Law will not save us from the resulting glut. The defining economic challenge of the coming decades will not be figuring out how to produce enough wealth, but engineering the systems required to distribute it.