Antitrust is where the machinery of market structure meets a decision with a deadline. It has three limbs.
Merger control reviews a transaction before it happens. Conduct enforcement attacks agreements between competitors, such as cartels and bid rigging, and unilateral abuse by a dominant firm, such as predatory pricing or refusal to supply. Market studies examine a sector without a specific defendant and often end in a recommendation for regulation rather than a penalty.
Merger review runs in a fixed order, and the order is the whole discipline.
How do you define the market and measure concentration?
Define the market first. A market is the smallest set of products and geographies over which a hypothetical single seller could profitably impose a small but significant non-transitory increase in price. The test is the SSNIP, and its arithmetic is the critical loss already derived in the market-structures work: with a tested price rise of t and a contribution margin m, the critical loss is CL = t / (t + m), and the candidate market stands if the actual loss, approximated by the market elasticity times t, is smaller.
Get this step wrong and every number after it is computed on the wrong table.
Measure concentration second. The Herfindahl-Hirschman Index is the sum of squared percentage shares. For a merger of two firms with shares s1 and s2, the change in the index has a closed form that is also your arithmetic check: ΔHHI = 2 × s1 × s2.
Compute it by subtraction and by the shortcut. If the two disagree, your share table does not sum to 100.
Advertisement
How do you screen for unilateral effects and entry?
Screen for unilateral effects third. Concentration is a property of the market; the harm from a specific merger is a property of the two merging firms' closeness to each other. The standard screen is the gross upward pricing pressure index, which asks what the merged firm now internalises. If a share of D of the customers who leave firm 1 after a price rise go to firm 2, and firm 2 earns a margin m2 on them, then firm 1's incentive to raise price has increased by GUPPI = D × m2 × (P2/P1).
Two firms with modest shares in a market full of close substitutes produce a small ΔHHI and a large GUPPI, and the second is the one that predicts harm.
Assess entry and countervailing power fourth. A high index in a market where a new plant can be built in eighteen months is a different object from the same index behind a scarce licence, and different again where three buyers each take a quarter of output.
Structural vs behavioural remedies
Then decide the remedy, and the choice between the two kinds is the most consequential judgment in the process. A structural remedy changes who owns what: divest a plant, a brand, a route, to a buyer capable of competing. It is self-executing, verifiable at closing, and once done it needs no supervision.
A behavioural remedy changes what the merged firm may do: a price cap, an access commitment, a non-discrimination undertaking, a firewall between divisions. It is cheaper for the parties, it can preserve genuine efficiencies that a divestiture would destroy, and it requires a monitor for its entire life, which means its value equals the probability that it is actually enforced.
The arithmetic to hold on to is uncomfortable and correct. A behavioural remedy changes neither the concentration index nor the SSNIP answer, because both of those are structural and definitional instruments and a promise is neither.
Worked Example 1 computes exactly this: the same merger, the same index of 2,653, the same market definition, and a remedy whose entire value is a probability.