What Happens When the System Is Nudged

The subject here is what happens when a policy or a changing investment market alters the thresholds that connect advanced technology, venture capital, and startup growth. The focus is not on finding a universal winner between VC and bank finance. It is on tracing how a shock changes which firms enter, adopt technology, receive funding, survive, and grow—and then comparing that mechanism with what the global AI-VC market actually measures. This distinction helps readers interpret GDP, employment, welfare, and investment headlines without assuming that a bigger number on one margin means a better outcome everywhere else.

The calibrated model is particularly useful for this purpose because it contains reallocation. A tax or subsidy does not simply make every existing firm more productive by the same amount. It changes the relative payoff of different projects, so some firms expand, some switch technology or finance, and some no longer cross the threshold needed to enter or survive. The OECD's global data show a similar need for caution at a different level: total AI investment can surge while value becomes concentrated in a few countries, stages, industries, and exceptionally large deals.

Counterfactuals move thresholds, not just productivity

The first counterfactual makes venture capital prohibitively expensive. In the model, eliminating VC reduces GDP by 4.4% and welfare-equivalent consumption by 3.0%. The loss reflects the disappearance of the modeled VC synergy: some projects no longer receive the financing relationship that raises their expected productivity. Yet high-tech adoption rises by 4.5%. That apparently paradoxical result comes from reallocation. Some projects that had been non-high-tech and VC-financed switch to high-tech and bank finance in an effort to preserve profitability. The technology label becomes more common, but the average suitability and productivity of the high-tech group fall when VC's value-added channel is removed.

The second counterfactual makes advanced technology prohibitively expensive. Its modeled effect is larger: GDP falls 12.9%, employment falls 8.2%, measured total factor productivity falls 5.5%, and welfare-equivalent consumption falls 9.0%. The comparison does not mean VC is unimportant. It separates reach from intensity. Advanced technology is used by a much larger share of firms, so removing it affects a broad base of production. VC-backed firms are exceptionally large per firm, but they are much fewer. Removing VC therefore damages a concentrated high-growth tail, while removing high tech removes a capability with wider coverage.

flowchart TD A["Tax, subsidy, or capital-market shock"] --> B["Financing and technology thresholds"] B --> C["Entry, exit, and firm-type mix"] C --> D["Wages and resource allocation"] D --> E["GDP, employment, TFP, and welfare"] F["Global AI VC: concentration and cyclicality"] -. "Context, not the same measure" .-> E

Taxes create another threshold effect. The experiment raises the corporate income tax and capital-gains tax to 28%, while partnerships face a 30% personal income tax with a 20% qualified-business-income deduction. It also assumes that all VC-funded startups are C corporations and all bank-funded startups are partnerships, with tax receipts rebated to consumers. Those organizational and fiscal assumptions are part of the model experiment; the resulting numbers are not estimates of what a real tax bill would do under every business form.