Lerner index¶

Collapse a firm's market power into one dimensionless number, the markup of price over marginal cost as a fraction of price, L = (P − MC) / P, which under profit maximization also equals the reciprocal of the demand elasticity the firm faces.

Core Idea¶

The Lerner index, introduced by Abba Lerner in 1934, is a scalar measure of a firm's market power defined as L = (P − MC) / P, the markup of price over marginal cost expressed as a fraction of price. It ranges from zero — indicating perfect competition, where price equals marginal cost and the firm earns no markup — to one, indicating a pure monopolist facing perfectly inelastic demand, where price could in principle be set arbitrarily above cost. For a profit-maximising firm, the index equals the reciprocal of the absolute price elasticity of demand it faces: L = 1/|ε|. This identity means that a firm's observed price–cost margin directly reveals the demand elasticity facing it, and conversely, knowing the elasticity predicts the markup the firm will charge under profit maximisation.

The index measures behavioural market power — the actual pricing deviation from the competitive benchmark — rather than structural concentration. A firm with a large market share in a competitive industry with many close substitutes may have a Lerner index near zero; a small firm with a differentiated product and few substitutes may have a Lerner index well above zero. The index is dimensionless and scale-independent, making it comparable across industries with very different absolute price levels: a Lerner index of 0.3 in banking means the same magnitude of markup relative to competitive pricing as 0.3 in hospital services, regardless of the difference in dollar prices.

In applied industrial organisation, the index serves two main roles. First, as a direct measurement instrument: using cost-accounting data or econometric cost estimation, analysts compute the price–cost margin for a firm or industry and use it to benchmark the degree of competitive pressure. Second, through the elasticity identity, it connects observable pricing behaviour to the unobservable demand elasticity the firm faces, supporting the New Empirical Industrial Organisation programme of inferring market conduct from prices and quantities. Empirical studies of banking competition, hospital pricing, and other concentrated industries routinely report industry-average Lerner indices as summary statistics of competitive intensity, with values above 0.3–0.4 typically indicating materially imperfect competition.

Structural Signature¶

Sig role-phrases:

the firm facing residual demand — a profit-maximiser facing a downward-sloping residual demand curve
the price and the marginal cost — the two quantities, both measurable in money per unit, whose gap is the raw material
the competitive-floor and monopoly anchors — the two fixed reference points: L = 0 at perfect competition (P = MC), L → 1 at pure monopoly facing inelastic demand
the normalized markup ratio — the construct itself, L = (P − MC) / P, the price-cost margin as a dimensionless, scale-free fraction of price
the scale-free comparability — the dimensionlessness that makes 0.3 in banking denote the same departure from competitive pricing as 0.3 in hospital services
the elasticity identity (its engineered bridge) — L = 1/|ε|, the profit-maximisation identity letting the observable markup and the unobservable demand elasticity each be inferred from the other
the behavioural-not-structural focus — the index measures power as exercised (the actual price-cost deviation), distinct from concentration measures like the Herfindahl-Hirschman index that measure structure
the normalized-slack look-alike (its limitation) — the same gap-to-floor-over-value form recurs in unrelated systems (e.g. CPU thermal headroom) only as a mathematical coincidence, absent the firm, residual demand, and elasticity identity that give L its content

What It Is Not¶

Not a measure of market concentration. The Lerner index measures power as exercised — the actual price-cost deviation — not market structure. Concentration measures (market share, the Herfindahl-Hirschman index) answer "how few firms are there?"; the Lerner index answers "how far above its competitive floor is this firm pricing?" The two need not move together: a dominant-share firm hemmed in by substitutes can have L near zero, and a small differentiated firm can have L high.
Not a causal mechanism. L = (P − MC) / P is a constructed ratio, and L = 1/|ε| is an identity under profit maximization, not a claim that elasticity causes the markup or vice versa. It is a window that lets the observable margin and the unobservable elasticity each be read from the other, not a force that sets prices.
Not proof of monopoly abuse or inefficiency. A high Lerner index records a markup; it does not by itself establish wrongdoing or even welfare loss. The same markup can reflect product differentiation, scarce substitutes, or recovery of large fixed costs (where pricing at marginal cost would mean losses), so L is a diagnostic to be interpreted, not a verdict.
Not the absolute markup. The index is a dimensionless, scale-free fraction of price, deliberately normalized so that 0.3 in banking denotes the same departure from competitive pricing as 0.3 in hospital services. Reading it as a dollar margin discards exactly the cross-industry comparability that is its point.
Not a general "normalized-slack" formula. The same gap-to-floor-over-value shape appears in unrelated systems (a CPU's thermal headroom), but those are mathematical coincidences with no profit-maximizing firm, no residual demand, and no elasticity identity behind them. What travels cross-domain is the generic gap-to-benchmark move, not "the Lerner index," whose microeconomic apparatus stays home.

Scope of Application¶

Because the Lerner index is a constructed measure — a normalized markup ratio L = (P − MC) / P carrying the identity L = 1/|ε| — and not a causal mechanism, its reach is set by a precondition rather than a domain boundary: it is computable, and means what it should, wherever a profit-maximising firm prices against downward-sloping residual demand with a price and a marginal cost both measurable in money per unit. That precondition happens to instantiate inside economics and the immediately adjacent quantitative-policy literature, so the fields below are genuine literal uses of the identical construct; the boundary to police is instrument-reach versus over-reading the bare ratio where its inputs do not carry the meaning the index needs (the "normalized-slack" look-alikes belong to the generic gap-to-benchmark move, not here).

Industrial organization and antitrust screening — the home use; the price-cost margin computed from cost data benchmarks competitive pressure and flags firms whose markup clears the imperfect-competition threshold (typically ~0.3–0.4) for closer scrutiny, distinct from the structural HHI reading.
Regulated-industry rate-setting — regulators use the markup over marginal cost to gauge how far a regulated firm prices above its competitive floor when assessing tariffs and allowed returns.
Banking-competition empirical literature — industry-aggregated Lerner indices are computed from banks' financial statements as summary statistics of competitive intensity (low in developed competitive markets, rising where a few national banks dominate).
Health-care and hospital pricing studies — regional inpatient Lerner indices measure provider market power and, where they correlate with market HHI, both validate the index against the structural measure and display the behavioural/structural complementarity.
New Empirical Industrial Organization — through the L = 1/|ε| identity the observable markup doubles as a statement about the residual-demand elasticity, letting analysts back out market conduct from prices and quantities without estimating a full demand system.

Clarity¶

The sharpest distinction the Lerner index makes legible is between market power as exercised and market structure as concentrated. Industrial organisation routinely reaches for concentration measures — market share, the Herfindahl-Hirschman index — as proxies for power, but the Lerner index measures the thing itself: the actual deviation of price from marginal cost. Once the two are separated, the analyst can name cases the concentration proxies misread — a dominant-share firm hemmed in by close substitutes and priced near cost (L near zero), or a small firm with a differentiated product and few rivals pricing well above cost (L high). The question shifts from "how few firms are there?" to "how far is this firm actually pricing above its competitive floor?" — and those need not move together.

The index also makes power comparable where dollar prices make it incomparable. Because L is a dimensionless fraction of price, a 0.3 markup in banking denotes the same departure from competitive pricing as 0.3 in hospital services, despite price levels differing by orders of magnitude — so cross-industry benchmarking becomes a single-number comparison rather than an apples-to-oranges one. Its second clarifying service is the identity L = 1/|ε|, which turns an observable — the price-cost margin read off cost data — into a window on an unobservable, the demand elasticity the firm faces. That equivalence is what lets the empirical-IO analyst infer the firm's competitive conduct from prices and quantities alone, and it sharpens the reading of any computed markup: a high Lerner index is, under profit maximisation, simultaneously a statement that the firm faces inelastic residual demand.

Manages Complexity¶

A firm's competitive situation is a thick dossier — its market share, the number and closeness of its rivals, the substitutes a buyer can turn to, product differentiation, entry barriers, the residual demand it actually faces, the absolute price level of its industry — and assessing market power across firms means reconciling all of that, in different currencies and at different price scales, into a verdict on how far each prices above its competitive floor. The Lerner index compresses that dossier into a single dimensionless scalar, L = (P − MC) / P, anchored at two fixed reference points (0 at perfect competition, approaching 1 at pure monopoly facing inelastic demand). The whole structural sprawl is replaced by one number an analyst reads directly off cost data: a markup above roughly 0.3-0.4 flags materially imperfect competition, and because the ratio is scale-free, 0.3 in banking and 0.3 in hospital services denote the same departure from competitive pricing despite price levels differing by orders of magnitude — so cross-industry power becomes a one-number comparison rather than an apples-to-oranges reconciliation. For first-cut antitrust screening or regulatory benchmarking, this single scalar stands in for the market-structure file.

Two features make the compression sharper than a mere summary statistic. First, it isolates the right quantity: the index measures power as exercised (the actual price-cost deviation) rather than structure as concentrated (share, the Herfindahl-Hirschman index), so an analyst tracking L correctly separates a dominant-share firm priced near cost by close substitutes (L near zero) from a small differentiated firm pricing well above cost (L high) — cases the concentration proxies misread. The analyst tracks one behavioural number instead of inferring power from a structural bundle that need not move with it. Second, the identity L = 1/|ε| collapses two quantities into one: the index that is read off observable prices and costs is simultaneously a statement about the unobservable demand elasticity the firm faces, so computing the markup and inferring the firm's competitive conduct become a single act rather than two. That equivalence is exactly what lets the empirical-IO analyst back out market conduct from prices and quantities alone — a high Lerner index reads at once as a large markup and as inelastic residual demand, two faces of one parameter.

Abstract Reasoning¶

The Lerner index licenses reasoning that reads market power off a single scalar and, through one identity, runs that reasoning in both directions between an observable markup and an unobservable elasticity.

The foundational move is measuring power as exercised rather than inferring it from structure. To judge a firm's market power, the analyst computes L = (P − MC) / P from cost data and reads the actual deviation of price from the competitive floor directly, rather than proxying power from concentration measures (market share, the Herfindahl-Hirschman index). The reasoning runs from an observed price-cost margin to a verdict on competitive pressure, anchored at two fixed reference points (0 at perfect competition, approaching 1 at pure monopoly facing inelastic demand): a markup above roughly 0.3-0.4 flags materially imperfect competition. This is the move that lets the analyst diagnose cases the concentration proxies misread — a dominant-share firm priced near cost by close substitutes (L near zero), or a small differentiated firm pricing well above cost (L high) — by tracking the behavioural number instead of the structural bundle that need not move with it.

The decisive move is bidirectional inference through the identity L = 1/|ε|. Because a profit-maximising firm's markup equals the reciprocal of the absolute elasticity of the residual demand it faces, the analyst reasons across the observable/unobservable boundary in either direction. Forward: a computed Lerner index is simultaneously a statement that the firm faces inelastic residual demand, so reading a high markup off cost data is reading the demand curve's elasticity at the same time — the foundation of the New Empirical Industrial Organisation move of backing out market conduct from prices and quantities alone, without a full demand system. Backward: knowing the elasticity a firm faces predicts the markup it will set under profit maximisation, so the analyst forecasts pricing behaviour from an elasticity estimate. The reasoning treats one parameter as having two faces and infers whichever face is hidden from whichever is observed.

A third move is scale-free cross-industry comparison. Because L is a dimensionless fraction of price, the analyst reasons that a 0.3 markup in banking denotes the same departure from competitive pricing as 0.3 in hospital services, despite price levels differing by orders of magnitude — so power becomes a single-number comparison across industries rather than an apples-to-oranges reconciliation. The move is to normalise away the absolute price scale and reason about relative deviation from the competitive benchmark, which is what makes industry-average Lerner indices usable as comparable summary statistics of competitive intensity.

A fourth move is boundary-drawing against structural measures, with a known complementarity. The analyst reasons explicitly about which question a given instrument answers — "how far is this firm pricing above its competitive floor?" (Lerner, behavioural) versus "how few firms are there?" (HHI, structural) — and treats the two as separately informative rather than interchangeable. This licenses both a contrast (the Lerner index can diverge from concentration when substitutes or differentiation break the share-to-power link) and a validation move (where regional Lerner indices for a service correlate with market HHI, the behavioural and structural readings corroborate each other), so the analyst knows when concentration is a safe proxy for power and when it is not.

Finally, the index supports a first-cut screening move that the single scalar makes possible. For antitrust triage or regulatory benchmarking, the analyst reasons from the one number to a provisional verdict — flag firms or industries whose markup clears the imperfect-competition threshold for closer scrutiny — letting L stand in for the full market-structure file at the screening stage, with the elasticity identity available to deepen the analysis into conduct inference where a flagged case warrants it.

Knowledge Transfer¶

The Lerner index is a constructed measure — a normalized markup ratio, L = (P − MC) / P, carrying an elasticity identity, not a causal mechanism — so "mechanism within / metaphor beyond" does not apply; what governs its travel is a precondition. The construct is computable, and means what it is supposed to mean (market power as exercised, plus the demand elasticity facing the firm), exactly where its ingredients exist: a firm facing downward-sloping residual demand, with a price and a marginal cost both measurable in money per unit, behaving as a profit-maximiser so that the L = 1/|ε| identity holds. Wherever that setup obtains the index transfers literally and as itself, anchored at its two fixed reference points (0 at perfect competition, approaching 1 at pure monopoly facing inelastic demand). The boundary to police is therefore instrument-reach versus over-reading — whether the residual-demand, profit-maximising setup is genuinely present, or whether the bare ratio is being computed where its inputs do not mean what the index needs them to.

Within economics and the immediately adjacent quantitative-policy literature the construct travels broadly on exactly that footing. The same scalar, the same scale-free cross-industry comparability (a 0.3 markup in banking denoting the same departure from competitive pricing as 0.3 in hospital services), the same behavioural-versus-structural contrast with the Herfindahl-Hirschman index, and the same bidirectional inference through L = 1/|ε| carry without translation across industrial organization and antitrust screening, regulated-industry rate-setting, the banking-competition empirical literature (industry-aggregated Lerner indices computed from financial statements), health-care pricing studies (regional inpatient Lerner indices correlating with market HHI, which both validates the index and demonstrates the complementarity), and the New Empirical Industrial Organization program of backing market conduct out of prices and quantities without a full demand system. Across all of these the construct is the same measure, computed the same way, meaning the same thing, wherever a profit-maximising firm prices against residual demand. This is instrument transfer in the strict sense.

Beyond economics the index does not transfer, and honesty requires marking the seductive look-alike. The same algebraic form — a gap-to-floor expressed as a fraction of value, "normalised slack" — reappears in unrelated places (a CPU's clock-speed headroom against its thermal limit can be written in this shape), but these are mathematical coincidences of normalized slack, not transfers of the Lerner mechanism, because there is no profit-maximising firm, no residual demand, and no elasticity identity behind them. Writing such a ratio and calling it "a Lerner index" is over-reading the formula past the point where its inputs mean what they should. What genuinely does generalize — and what those look-alikes are really instances of — is not the Lerner index but the generic move it instantiates: measure deviation from a theoretical benchmark as a normalized, scale-free ratio. That move is already carried, in the market-power setting, by competition, price_mechanism, and markup_pricing, and more generally by the efficiency_frontier-style gap-to-benchmark normalization that recurs across disciplines (and would sit naturally alongside the Gini, Theil, HHI, and Sharpe ratios in any cluster of diagnostic indices). Strip the microeconomic apparatus — the firm, residual demand, marginal cost, the elasticity identity — and what remains is exactly that generic normalized-gap move, not the Lerner index. So the honest split is between instrument-reach (the measure is computable, and informative, wherever a profit-maximising firm faces residual demand with measurable price and marginal cost — which is why the IO, antitrust, banking, and health-care uses are co-instances) and over-reading (the normalized-slack ratio appears in physical and other systems only as a formal coincidence, and the portable content there is the generic gap-to-benchmark move, not "the Lerner index"). The full split is drawn in Structural Core vs. Domain Accent.

Examples¶

Canonical¶

Take a firm selling at a price of $10 per unit whose marginal cost of producing one more unit is $6. The Lerner index is L = (P − MC) / P = (10 − 6) / 10 = 4/10 = 0.4 — a markup of 40% of price above the competitive floor, comfortably past the ~0.3–0.4 band that flags materially imperfect competition. The elasticity identity then reads the demand side off the same number: under profit maximization L = 1/|ε|, so |ε| = 1/L = 1/0.4 = 2.5. The firm is behaving as though the residual demand it faces has an absolute price elasticity of 2.5. Had marginal cost instead equalled the $10 price, L would be 0 — the perfect-competition anchor, with implied elasticity infinite (perfectly elastic demand). The single arithmetic step delivers both the exercised markup and the elasticity of the demand curve the firm faces.

Mapped back: The $10 and $6 are the price and the marginal cost whose gap feeds the normalized markup ratio L = 0.4. Reading |ε| = 2.5 straight off L is the elasticity identity, the engineered bridge letting the observable margin reveal the unobservable elasticity. L = 0 when P = MC is the competitive-floor anchor. The number captures power as exercised, not concentration — the behavioural-not-structural focus.

Applied / In Practice¶

Banking-competition research computes industry-average Lerner indices directly from banks' financial statements: price is proxied by total revenue over total assets, and marginal cost is estimated from a translog cost function fitted to the same accounts. The World Bank's Global Financial Development Database reports these country-level banking Lerner indices as standard summary statistics of competitive intensity, and the empirical literature (e.g. work following Demirgüç-Kunt and Martínez Pería) uses them to compare market power across banking systems. A low index signals a competitive banking sector; a higher one signals that a few dominant banks price well above marginal cost. Because the index is dimensionless, a 0.30 reading in one country's banking system is directly comparable to 0.30 in another's, despite different currencies and price levels.

Mapped back: Computing L from revenue-over-assets and an estimated cost function is the normalized markup ratio applied to real accounts, capturing power as exercised rather than the structural concentration of the sector. Comparing 0.30 across countries with different currencies is the scale-free comparability — the dimensionlessness that makes cross-system benchmarking a one-number comparison.

Structural Tensions¶

T1: Power as exercised versus structure as concentrated (the instrument fixes which question you can answer). The Lerner index deliberately measures behavioural market power — the actual price-cost deviation — rather than structural concentration, and this is its signature virtue: it correctly reads a dominant-share firm hemmed in by substitutes as near-competitive (L≈0) and a small differentiated firm as powerful (L high), cases the HHI misreads. But the choice is not free of cost. The behavioural number tells you that a firm prices above its floor without telling you why in structural terms, while concentration tells you the structure without the exercised markup. The two are complementary, and where they corroborate (regional Lerner correlating with HHI) each validates the other — but where they diverge, the analyst must decide which reading governs, and the Lerner index alone cannot say whether the markup reflects a structural problem or a benign one. Diagnostic: Is the question here "how far is this firm actually pricing above cost?" (Lerner) or "what market structure produced that power?" (HHI) — and do the behavioural and structural readings agree or diverge?

T2: A markup measured versus a verdict inferred (a high L flags but does not convict). The single scalar makes first-cut screening possible: a markup past ~0.3–0.4 flags materially imperfect competition. But the compression that enables triage discards exactly the context needed to judge the flag. The same high L can reflect monopoly abuse, or product differentiation buyers genuinely value, or the recovery of large fixed costs where pricing at marginal cost would mean losses. The tension is that the index records a markup and is routinely read as a verdict on wrongdoing or welfare loss, when it is a diagnostic to be interpreted — the number that makes the screen fast is the number stripped of the information that would tell you whether the markup is a problem. Diagnostic: Does the high Lerner index here indicate exercised power that harms welfare, or a markup explained by differentiation or fixed-cost recovery that marginal-cost pricing could not sustain?

T3: Ease of computation versus the fragility of measuring marginal cost (the hard input hides in plain sight). "Read L off cost data" presents the index as directly observable, and that apparent computability is much of its appeal. But marginal cost — the denominator's silent partner — is one of the hardest quantities in applied economics to measure: accounting cost is not marginal cost, and estimates lean on translog cost functions, allocation assumptions, and functional-form choices that can move the computed L materially. The tension is that the index's reputation as a clean, read-off-the-books number conceals that its key input is an econometric construct laden with assumptions, so two analysts with the same accounts can report different Lerner indices depending on how they estimated MC. Diagnostic: Is the marginal cost in this computation genuinely marginal, or an accounting or estimated proxy whose assumptions could shift the index across the imperfect-competition threshold?

T4: The elasticity identity's reach versus its profit-maximization premise (backing out conduct assumes the conduct). The identity L = 1/|ε| is the index's most powerful move — it lets the observable markup double as a statement about the unobservable demand elasticity, the engine of the New Empirical IO program's inference of conduct from prices and quantities. But the identity holds only under profit maximization, so using L to infer the elasticity the firm faces presupposes exactly the optimizing conduct that a conduct analysis might want to test. The tension is that the bridge between markup and elasticity is not a fact but a behavioral assumption, so inferences that ride across it are conditional on the firm behaving as a static profit-maximizer — and a firm pricing under other objectives (market-share, regulation, collusion tacit or overt) breaks the identity the inference depends on. Diagnostic: Is the firm here a static profit-maximizer (so L = 1/|ε| licenses the elasticity read), or is its pricing governed by objectives that void the identity the conduct inference rests on?

T5: A static markup versus dynamic competition (a snapshot can misread a contestable market). The Lerner index is a point-in-time measure of exercised power, and a high value reads as imperfect competition. But intense dynamic competition can coexist with high current markups: a contestable market disciplined by the threat of entry, or a Schumpeterian innovator earning temporary rents that fund and are eroded by the next innovation, can show a high instantaneous L while being fiercely competitive over time. The tension is that the index captures the current price-cost gap but is blind to the potential and dynamic competition that may be constraining it, so a snapshot markup can misclassify a market that is competitive in the dimension that matters (entry, innovation) as monopolistic. Diagnostic: Is the high current markup here evidence of durable market power, or a transient rent in a contestable or innovation-driven market that dynamic competition is already disciplining?

T6: Autonomy versus reduction (a named IO measure or the generic gap-to-benchmark normalization). The Lerner index is a specific, canonically defined construct — the normalized markup L = (P − MC)/P with the identity L = 1/|ε| — and it transfers literally, as itself, wherever its precondition holds: a profit-maximizing firm pricing against downward-sloping residual demand with measurable price and marginal cost. That is why the IO, antitrust, banking, and health-care uses are co-instances of the identical measure. But the same algebraic form — a gap-to-floor as a fraction of value — recurs in unrelated systems (CPU thermal headroom) only as a mathematical coincidence, with no firm, no residual demand, no elasticity identity. What genuinely generalizes is the generic move the index instantiates: measure deviation from a theoretical benchmark as a normalized, scale-free ratio, already carried by competition, price_mechanism, markup_pricing, and the efficiency_frontier-style gap-to-benchmark normalization. Diagnostic: Resolve toward the generic normalized-gap move (and markup_pricing/competition) when the setting lacks a profit-maximizing firm and residual demand; toward the Lerner index when those inputs are present and mean what the identity requires.

Structural–Framed Character¶

The Lerner index sits at mixed on the structural–framed spectrum — a constructed economic measure with a clean mathematical core (a normalized ratio plus an optimization identity), but one invented by an economist and defined entirely over human-economic constructs, which pulls it toward the framed side.

Evaluative weight is low and points structural. The index is a diagnostic ratio, not a verdict; the entry is emphatic that a high L "records a markup" without establishing wrongdoing or welfare loss. Its antitrust use gives it evaluative stakes, but the measure itself is a neutral scalar, and the concept explicitly refuses to convict.

Human-practice-bound pulls toward framed, though not by observer-dependence. A firm's markup is what it is unmeasured (not constituted by the analyst), but the entire subject matter — firms, prices, marginal cost, profit maximization, residual demand, elasticity — is the apparatus of a market economy and of microeconomic theory. There is no Lerner index in nature; it exists only where a constructed economic setup obtains. So it is practice-bound at the substrate level: its inputs are artifacts of markets and of the theory that models them.

Institutional origin is mixed, tilting framed: the Lerner index is a constructed instrument — named after Abba Lerner (1934), invented by the discipline of industrial organization, computed by convention (the ~0.3–0.4 threshold, the translog cost estimation) — which is the artifact-of-a-theory character that reads framed. But the underlying content — a normalized gap-to-benchmark, and the identity L = 1/|ε| as a mathematical consequence of optimization — is not an institution; it is a construct with a genuine mathematical backbone.

Vocab-travels is bounded: markup, marginal cost, residual demand, elasticity are economics-bound, so the named index travels literally only where its precondition holds and becomes a mere formal coincidence (CPU thermal headroom) elsewhere. Import-vs-recognize is bimodal in the entry's own terms: within economics the identical measure transfers as a co-instance (recognition), while cross-domain look-alikes are the generic normalized-gap move, not the Lerner index imported.

The portable structural skeleton is deviation from a theoretical benchmark expressed as a normalized, scale-free ratio — carried by the efficiency_frontier-style gap-to-benchmark normalization generally, and by markup_pricing, competition, and price_mechanism in the market-power setting. As the entry establishes, that skeleton is what the Lerner index instantiates and specializes, not what makes "the Lerner index" itself travel: the cross-domain reach belongs to that generic normalized-gap move (kin to the Gini, Theil, HHI, and Sharpe ratios), while the domain-accented apparatus — the firm, residual demand, marginal cost, and the engineered elasticity identity L = 1/|ε| — stays home. Its character: an evaluatively neutral, mathematically clean normalized-gap measure invented by economists and defined over market constructs, structural in its gap-to-benchmark core but pinned by its constructed-instrument origin and economic substrate to mixed rather than a free-floating prime.

Structural Core vs. Domain Accent¶

This is the section that decides why the Lerner index is a domain-specific abstraction and not a prime — a case with its own twist, because the index is a constructed measure rather than a mechanism, so its home-versus-away boundary is drawn as instrument-reach versus formal coincidence rather than mechanism versus metaphor.

What is skeletal (could lift toward a cross-domain prime). Strip the microeconomics and a thin abstract structure survives: the deviation of an observed value from a theoretical benchmark, expressed as a normalized, scale-free ratio of the gap to the value, anchored at fixed reference points. The portable pieces are abstract: a benchmark (the competitive floor), a measured quantity above it, a gap, and a dimensionless normalization that makes the ratio comparable across settings of wildly different absolute scale. That skeleton is genuinely substrate-portable — it is the generic gap-to-benchmark move that recurs across diagnostic indices (kin to the Gini, Theil, HHI, and Sharpe ratios) — which is why what actually generalizes is carried by the efficiency_frontier-style gap-to-benchmark normalization in general, and by markup_pricing, competition, and price_mechanism in the specific market-power setting. But it is the generic move the index instantiates and specializes, not what makes "the Lerner index" itself distinctive.

What is domain-bound. Everything that makes the construct the Lerner index in particular is microeconomic apparatus that does not survive extraction. The benchmark is specifically the perfect-competition floor where price equals marginal cost; the measured quantity is a firm's price against its marginal cost; the anchors are L = 0 at perfect competition and L → 1 at pure monopoly facing inelastic demand; and the load-bearing addition is the engineered elasticity identity L = 1/|ε|, which holds only for a profit-maximiser facing downward-sloping residual demand and lets the observable markup and the unobservable demand elasticity each be read off the other. The empirical instances — banking Lerner indices from translog cost functions, regional hospital-pricing indices correlating with HHI, the New Empirical IO conduct inference — are all built from this apparatus. The decisive test: remove the profit-maximising firm and its residual demand, and the same gap-to-floor-over-value ratio can still be written (a CPU's thermal headroom fits the algebra), but it is a mathematical coincidence with no elasticity identity and no market-power content — no longer the Lerner index, just the generic normalized-slack shape.

Why this does not clear the prime bar. A prime is a relational structure whose vocabulary travels and whose cross-domain transfer is recognition of the same thing, not analogy or formal accident. The Lerner index's reach is bimodal in exactly the way a constructed measure's is. Within economics and the adjacent quantitative-policy literature the identical construct transfers literally, as itself, wherever its precondition holds — antitrust screening, regulated rate-setting, the banking-competition literature, health-care pricing studies, and the New Empirical IO program all compute the same scalar the same way and mean the same thing, because each supplies a profit-maximising firm pricing against residual demand with measurable price and marginal cost. Beyond that precondition the "transfer" is only the recurrence of the algebraic form: writing a gap-to-floor ratio in a thermal or other physical system and calling it a Lerner index is over-reading the formula past the point where its inputs mean what the identity requires. And when the bare structural lesson — measure deviation from a benchmark as a normalized, scale-free ratio — is needed cross-domain, it is already carried, in fully general form, by that generic gap-to-benchmark normalization (with markup_pricing, competition, and price_mechanism for the market-power case specifically). The cross-domain reach belongs to that generic move; "the Lerner index," as named, stays home with the firm, residual demand, marginal cost, and the engineered L = 1/|ε| identity that give the scalar its economic content.

Relationships to Other Abstractions¶

Current abstraction Lerner index Domain-specific

Parents (3) — more general patterns this builds on

Lerner index presupposes Market power Domain-specific

The Lerner Index presupposes market power as the target attribute whose exercised price-cost wedge it maps onto a scale.
Lerner index is part of, conditional Elasticity Prime

Under interior profit maximization, the index contains the reciprocal residual-demand elasticity identity that makes markup and responsiveness two readings of one parameter.
Lerner index is a decomposition of Measurement Prime

Removing firm and price vocabulary leaves an attribute mapped by a stated procedure onto a dimensionless scale with fixed anchors and interpretation conditions.

Hierarchy paths (4) — routes to 3 parentless roots

Lerner index → Market power → Bargaining Power → Asymmetry

Show alternative paths (3)

Not to Be Confused With¶

Herfindahl-Hirschman index (HHI). The standard structural concentration measure — the sum of squared market shares — answering "how few firms are there?" The Lerner index measures power as exercised (the actual price-cost markup), answering "how far above its competitive floor is this firm pricing?" They can diverge: a dominant-share firm hemmed in by substitutes has high HHI but L near zero. Tell: is the number built from market shares/structure (HHI), or from price minus marginal cost (Lerner)? Where they correlate they corroborate; where they diverge they answer different questions.
Absolute markup / price-cost margin (in dollars). The raw money gap between price and marginal cost, P − MC. The Lerner index normalizes that gap by price, L = (P − MC)/P, yielding a dimensionless, scale-free fraction so 0.3 in banking equals 0.3 in hospital services regardless of price level. Tell: is the figure a dollar amount per unit (absolute markup), or a unitless fraction of price comparable across industries (Lerner index)? Reading L as a dollar margin discards its whole comparability point.
Price elasticity of demand. The responsiveness of quantity to price, |ε|. The Lerner index is linked to it by the identity L = 1/|ε| under profit maximization — but they are distinct quantities: one is a markup read off cost data, the other a property of the demand curve. The identity lets each be inferred from the other; it does not make them the same thing, and it holds only under profit-maximizing conduct. Tell: is the quantity a markup over cost (Lerner), or a demand-response ratio (elasticity)? The identity is a bridge, not an equation of identity between the concepts.
Sibling normalized diagnostic indices (Gini, Theil, Sharpe ratio). Other single-number measures built on the same gap-to-benchmark, scale-free template — inequality against perfect equality, return against risk. They share the Lerner index's form but measure entirely different referents with no firm, residual demand, or elasticity identity. Tell: does the ratio measure a firm's markup over its competitive floor (Lerner), or some other deviation-from-benchmark (inequality, risk-adjusted return)? Shared shape, different content.
The generic gap-to-benchmark normalization (+ markup pricing, competition, price mechanism). The substrate-neutral move — measure deviation from a theoretical benchmark as a normalized, scale-free ratio — that the Lerner index instantiates for market power. A CPU's thermal headroom written in the same algebra is this generic form, not a Lerner index (no firm, no elasticity identity). Tell: strip away the profit-maximizing firm, residual demand, and marginal cost and what remains is the bare normalized-gap move (carried by efficiency_frontier-style normalization, markup_pricing, competition), not the Lerner index. (Treated fully in an earlier section.)

Neighborhood in Abstraction Space¶

Lerner index sits in a crowded region of the domain-specific corpus (5^th percentile for distinctiveness): several abstractions share nearly its structure, so a description that fits it tends to fit its neighbors too.

Family — Market Structure & Price Equilibrium (25 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-07-12