Model notes How the economy works

Automation displacement compounds against the jobs that remain. Firms can also eliminate labour when consumer demand falls. The strength of that response is controlled by the demand sensitivity setting.

A sensitivity below 1 assumes that firms absorb part of the shock through margins, reduced hours, or other adjustments before eliminating jobs. Profit losses are absorbed through firm balance sheets rather than being modeled as an additional reduction in consumption.

The debt-funded UBI remains fixed even as displacement grows. Scenario four provides the indexed and funded alternative.

The agentic contribution is assessed against the wage-equivalent value of automated labour. Everything collected is passed through to households as transfers, so funding rises and falls with measured displacement by construction.

It is a spend-what-you-collect mechanism. Under-disclosure reduces the transfer households receive, but it cannot create a government deficit.

This single-jurisdiction model assumes that a dual-nexus allocation rule, combining activity nexus and consumer nexus, successfully anchors the agentic-income tax base. The simulator models the economic consequences of that assumption. It does not attempt to prove the legal effectiveness of the rule or model the cross-border apportionment process itself.

The indexed UBI separates the entitlement from its funding source.

The transfer replaces a selected share of the economy’s aggregate wage-income shortfall. Household income loss therefore determines how much the government owes. Financing comes separately through a non-deductible surtax on domestic corporate EBITA. The surtax rate adjusts to fund the entitlement, subject to a legislated maximum rate.

In this economy, operating surplus before levies is treated as EBITA because there is no interest or amortization to subtract.

The statutory surtax rate is increased to account for expected non-collection. This creates two possible regimes.

Below the legislated cap, the entitlement remains fully funded. Higher non-collection simply pushes the required statutory rate toward the cap more quickly. Once the required rate exceeds the cap, however, the available tax base can no longer finance the entitlement. The remaining gap becomes a structural deficit that no rate formula can legislate away.

The non-collection setting is a collection haircut, not a complete model of profit shifting. Cross-border tax structures, their effect on ordinary corporate tax revenue, and their effect on the domestic circular flow would require a multi-jurisdiction model.

The UBI index is measured against gross wages, while the transfer itself is untaxed. Because workers previously paid 20% income tax, replacing 70% of lost gross wages restores 87.5% of the lost disposable wage income.

The simulator can model one important difference between the two mechanisms: what happens when firms do not pay.

With the agentic contribution, underpayment reduces household transfers immediately, but no deficit opens. With the funded UBI, the entitlement remains intact. The avoided amount stays with the firms that avoided payment, while the funding gap moves onto the public balance sheet.

Under avoidance, the funded UBI does not fail its recipients. It fails its funders.

Other differences, including how each base would be measured in practice and whether the required rates would survive the political process, sit outside the circular-flow model.

The windfall chart separates three figures that are often conflated:

Profit before automation Profit after automation but before any levy Profit after the levy

The chart shows how much of the pre-levy automation windfall each mechanism leaves with firms. The assessment bases are not identical. The agentic contribution is based on automated wage-equivalent output, while the funded UBI surtax is applied to taxable EBITA.

Under the default settings, firms that pay the full agentic contribution remain more profitable than they were before automating. The contribution does not tax automation out of existence. It shares part of the productivity gain so that firms retain customers capable of buying what the automated economy produces.

The optional secondary-cost mechanism is one transparent equation, not a detailed model of any individual public system.

It combines a linear cost based on lagged employment and income shortfalls with a quadratic cost that begins once unemployment passes the selected stress threshold. The quadratic term represents the possibility that costs accelerate as pressure builds across supplemental income supports, housing, healthcare, retraining, public safety, and the justice system.

These costs are treated only as fiscal expenses. They are deliberately not recycled into consumer demand. Turning the mechanism on can therefore worsen a scenario’s budget, but it cannot worsen the simulated economy itself.

After any parameter change, the starting economy is renormalized so that every experiment begins with the same initial output. This allows scenarios to be compared from a common baseline rather than from differently sized economies.

The model does not include open-economy trade beyond foreign compute imports, new job creation, wage adjustment, monetary policy, capital gains channels, or any productivity dividend from AI. The wider series argues that excluding productivity gains makes the wage-equivalent figure a floor rather than a ceiling, but that claim is not modeled here.

A model this small cannot predict the future. It can only show that, when the economy is drawn this way, the loop does not close by itself.