Why Bans Can Misfire—and What a Tax Changes

This chapter examines what happens when policymakers try to regulate startup acquisitions after learning that some related deals can weaken idea implementation. The specific comparison is between banning related acquisitions and taxing related and unrelated acquisitions while allowing founders, incumbents, and investors to adjust. The value of the comparison is to show why a policy that improves one post-deal outcome can still reduce future entry or growth once the whole innovation system responds.

A ban has different effects when decisions are held fixed

The BSE model evaluates policy in an economy where start-ups create ideas, incumbents innovate on existing products, and both sides decide whether to search for and implement ideas. It begins with a no-policy balanced-growth path and then considers a surprise policy announcement followed by a transition to a new long-run equilibrium. This setup matters because an acquisition ban does not merely erase a transaction already on the table. It changes the expected payoff from creating a start-up, the threat incumbents face from entry, and the resources devoted to search and implementation.

The policy analysis tracks three innovation margins. The first is incumbent own innovation: improvements to existing products and innovations that allow an incumbent to displace another producer. The second is the startup rate, the number of new start-ups created. The third is the implementation rate of startup ideas, the share of ideas that are actually developed after they appear. The model’s growth decomposition gives incumbent own innovation roughly three-quarters of the weight—about 76% in the baseline. Entrants still matter, accounting for 21.1% of calibrated growth, but the margins do not contribute equally or move in the same direction.

Consumption adds another dimension. Resources used to create start-ups, search for them, and implement ideas cannot be consumed at the same time. A policy can therefore raise a measured innovation rate while lowering the share of output available for consumption, or reduce search waste while weakening future growth. The model evaluates the transition using consumption-equivalent welfare: the permanent shift in consumption in the no-policy economy that would make a person indifferent between staying on the old path and experiencing the policy transition. It is an aggregate measure, not a complete account of distributional or competitive consequences.

The first counterfactual holds most decisions fixed. The model bans related or “killer” acquisitions by setting their frequency to zero, but leaves the startup rate, incumbent own innovation, and unrelated-acquisition search unchanged. In that partial-equilibrium exercise, the implementation probability rises mechanically because a related incumbent can no longer acquire and under-implement the idea. Long-run growth rises by 4 basis points, and consumption-equivalent welfare rises by 1.29%. This is the intuitive case for a ban: remove the channel that suppresses implementation, and the direct innovation gain appears clearly.

But holding the other decisions fixed removes the feedback that made acquisition valuable in the first place. In the full model, a startup agrees to an acquisition because the price is above the value of remaining independent.