Price Discrimination

Definition and motivation

Price discrimination is any pricing policy that charges different prices to different buyers for essentially the same good or service when differences are not driven by cost differences. For price discrimination to be feasible the firm needs: (i) market power (ability to set price above marginal cost), (ii) the ability to segment markets or observe signals correlated with willingness to pay, and (iii) limited arbitrage (customers cannot easily resell or arbitrage away price differences).

Degrees of price discrimination (core taxonomy)

• •First-degree (perfect) price discrimination: the seller charges each consumer her exact willingness to pay for each unit. This extracts all consumer surplus and, if costs allow, produces the first-best allocation (output where marginal willingness to pay equals marginal cost). It requires extraordinary information — essentially the entire demand curve of each consumer.

• •Second-degree price discrimination: the seller offers a menu of quantity- or quality-contingent prices (nonlinear tariffs) and lets consumers self-select. Examples: block pricing, quantity discounts, versioning of products, two-part tariffs.

• •Third-degree price discrimination: the seller observes an observable characteristic (group membership) correlated with willingness to pay and sets different linear prices for each segment (student discounts, senior fares, geographic pricing).

Core intuition in 1 graph/formula

Start from a standard monopoly demand: $P(Q)=a-bQ$. In "In Profit Maximization, we learned" that monopoly chooses output where marginal revenue equals marginal cost: $MR(Q)=a-2bQ=MC$. Solve for monopoly quantity and price:

Concrete numeric example: let $a=100$, $b=1$, $MC=10$. Then

Consumer surplus (see "In Consumer Surplus, we learned") is the area between demand and price: here

Why discriminate? Because different customers have different maximum willingness to pay. If the monopolist could charge each consumer her exact willingness to pay (first-degree), output expands to the efficient level where $P(Q)=MC$, eliminating deadweight loss and capturing all surplus as profit. In practice, firms use weaker devices (second- and third-degree) to harvest part of that surplus.

When thinking about welfare: price discrimination reallocates surplus from consumers to producers. It can increase total welfare relative to single-price monopoly when discrimination leads to higher aggregate output (reducing deadweight loss), but it can reduce welfare if it lowers output in some segments even while increasing profit.

First-degree (perfect) price discrimination (PD)

Mechanics: with perfect information the firm charges each infinitesimal unit at the consumer's marginal willingness to pay, so output is chosen where $P(Q)=MC$. Using $P(Q)=a-bQ$, efficiency requires

Compare to single-price monopoly $Q_M=(a-MC)/(2b)$. The first-degree monopolist doubles output relative to single-price monopoly for a linear demand and captures the entire area under the demand curve up to $Q^*$ as profit.

Numeric example (continuing earlier): $a=100, b=1, MC=10$ gives

Total welfare under first-degree PD equals the competitive-efficient total surplus: producer profit equals the entire sum of areas under the demand curve minus total cost; consumer surplus is zero.

Third-degree PD (observable groups)

Suppose the monopolist serves two segments 1 and 2 with inverse demands

and common marginal cost $MC$. The firm maximizes profit

First-order conditions (see "In Profit Maximization, we learned") set marginal revenue in each market equal to marginal cost:

Therefore

Concrete numeric example: take $a_1=80,b_1=1$ (high-value market), $a_2=60,b_2=1$ (low-value market), and $MC=10$. Then

• •Market 1: $Q_1=(80-10)/2=35$, $P_1=(80+10)/2=45$.
• •Market 2: $Q_2=(60-10)/2=25$, $P_2=(60+10)/2=35$.

Compare single-price monopoly: aggregate demand is $P(Q)=a_{agg}-b_{agg}Q$ only if markets identical. With different intercepts we find the single optimal price by aggregating demand horizontally at each price. But commonly third-degree PD produces higher profit because the firm charges higher price where willingness-to-pay is higher.

Markup rule and elasticity

A compact result: the Lerner index generalizes to each segment:

where $\varepsilon_i$ is the price elasticity of demand in market $i$ evaluated at $P_i$. This follows from $MR_i=P_i\left(1+\frac{1}{\varepsilon_i}\right)=MC$. Example numeric: if $\varepsilon=-2$ at the chosen price, then the markup is $\frac{P-MC}{P}=0.5$. If $MC=10$ and elasticity -2, then $P=2MC=20$.

Welfare comparisons

• •First-degree PD: output is efficient, no deadweight loss; consumer surplus = 0, producer surplus maximized.
• •Third-degree PD: ambiguous effect on total welfare relative to single-price monopoly. If PD raises total output (for example, because single-price sets price driven by aggregate demand which may be high), DWL can fall; but PD can also lower output in some segments relative to single-price and increase DWL.

A useful rule: when demands across segments are differently elastic, segmenting allows the monopolist to price closer to MC in elastic segments (in absolute terms, smaller markup) and farther above MC in inelastic segments; this can raise total quantity if previously demand in elastic segment was suppressed by a high single price. Always compute aggregate output to compare DWL explicitly.

Concrete welfare calculation (from earlier numeric example): compute CS and PS in each market by area formulas and sum. If the segmented output sum exceeds the single-price output, total DWL falls relative to single-price monopoly; otherwise it may rise.

Second-degree PD (menues and self-selection)

Second-degree discrimination offers a menu of nonlinear prices so consumers self-select according to private type (willingness to pay). Typical devices: quantity discounts, block pricing, versioning, and two-part tariffs. To analyze, use mechanism design constraints: incentive compatibility (IC) and individual rationality (IR).

Two-part tariffs (TPT)

A two-part tariff is a price schedule consisting of a fixed fee $F$ plus a per-unit price $p$: a consumer buying $q$ pays $T(q)=F+pq$. For identical consumers with downward-sloping demand $P(q)$ and zero marginal cost, the profit-maximizing two-part tariff sets $p=MC$ to get efficient consumption and $F$ equal to the consumer surplus at that price to extract surplus entirely. This relies on "In Profit Maximization, we learned" that marginal price should be set by marginal conditions; and "In Consumer Surplus, we learned" how to compute the consumer surplus to set $F$.

Concrete numeric example: single consumer with inverse demand $P(q)=100-q$, $MC=0$. Set $p=MC=0$ so the consumer consumes $q^{*}=100$. Consumer surplus is

A monopolist sets $F=5000$ and obtains profit $\Pi=5000$ (and the per-unit margin is zero). Consumer surplus is zero; total output is efficient. Against heterogeneous consumers, one cannot extract full surplus from all types with a single $F$; the monopolist trades off participation of low-value types.

Example with two discrete types (numerical): suppose two consumers each have inverse demands $P_i(q)=a_i-q$, $MC=0$, and each consumer buys individually. Let $a_H=120,a_L=60,b=1$. If the monopolist sets $p=0$, then $q_H=120,q_L=60$, and their consumer surpluses are $CS_H=1/2\times120\times120=7200$, $CS_L=1/2\times60\times60=1800$. To sell to both, the maximum uniform fee is $F=1800$, so profit $=2\times1800=3600$. Alternatively, exclude the low type and charge $F=7200$ to the high type, profit $=7200$. So the firm may prefer to price so that only high-type participates. This illustrates the trade-off in two-part tariffs with heterogeneity.

Versioning and quality differentiation

Versioning sells multiple versions (qualities) of a product at different prices intended to segment consumers by willingness to pay. The canonical analysis (Mussa–Rosen) models consumers by a one-dimensional willingness-to-pay parameter $\theta$ and quality $q$. Utility for a type $\theta$ buying quality $q$ at price $p$ is $U(\theta,q,p)=\theta\,q-p$. Cost is $C(q)$ (often linear in $q$). The monopolist offers a menu $(q,p)$ to maximize profits subject to IC and IR constraints. A key result: the high-type gets an efficient quality choice (no distortion) while lower types are allocated a quality distorted downward to loosen IC constraints and extract surplus. This is a direct application of constrained profit maximization taught in "In Profit Maximization, we learned".

Concrete numeric illustration (sketch): Suppose two types: low $\theta_L=6$, high $\theta_H=10$, cost $C(q)=2q$, proportion of high types $\lambda=0.5$. The efficient quality for type $\theta$ solves $\theta=MC(q)$ where $MC(q)=2$, so both types would want positive quality if $\theta>2$. But the optimal contract solves a constrained optimization and typically yields $q_L$ lower than efficient because the monopolist must prevent high types from pretending to be low.

Bundling

Bundling is selling two (or more) goods together at a package price. Bundling is attractive when consumer valuations for goods are negatively correlated or when individual valuations have high variance: bundling reduces the variance of the total valuation across consumers, allowing a single package price to extract more surplus. For independent uniformly distributed valuations, pure bundling can raise profit versus separate pricing. Concrete numeric example: two goods A and B; consumers have independent uniform valuations on [0,100]. If sold separately, profit per good at monopoly price 50 yields expected profit 50*Pr(v>50) etc. Bundling sets package price near combined mean (100) and captures more of the mass above that price. Always compute expected revenue to compare.

Welfare implications of second-degree devices

• •Two-part tariffs with $p=MC$ are efficient in quantity but redistribute surplus from consumers to producers via the lump-sum fee.
• •Versioning and menus generally retain inefficiencies relative to first-best because of distortions imposed to satisfy IC constraints; welfare is often lower than first-best but higher firm profit.
• •Bundling may increase total output for some consumers (those who purchase a bundle but would not have bought each separate good), thus potentially reducing deadweight loss compared to single-good monopolies; effect depends on correlation of valuations.

Real-world applications (concrete examples)

• •Airlines: third-degree PD across observable segments (advance purchase, student, senior fares) and second-degree PD via fare classes (refundable vs non-refundable). Airlines use dynamic versioning (fare buckets) that approximate first-degree PD in rich data environments.

• •Software and cloud services: two-part tariffs (subscription fee + usage), versioning ("basic", "pro", "enterprise"), and bundling (office suites). For example, a SaaS firm often sets $p\approx MC$ for marginal usage and extracts consumer surplus via subscription fees.

• •Utilities and telecommunications: block tariffs (increasing block pricing, quantity discounts), off-peak/peak pricing (time-based PD), and connection fees (two-part tariffs).

• •Retail coupons and rebates: second-degree PD by letting consumers self-select into lower prices if they search/clip coupons; third-degree PD where discount code reveals group identity.

Policy and regulation implications

Antitrust and consumer protection: regulators are concerned when PD is used to price discriminate against protected groups or to engage in price squeezing and exclusion. Price parity clauses (e.g., hotels and online platforms) restrict a platform’s ability to employ different prices on different channels. First-degree PD often raises distributional concerns despite improving allocative efficiency (it eliminates consumer surplus), so policy weighs equity versus efficiency.

Welfare nuances to remember

• •First-degree PD achieves allocative efficiency but transfers all surplus to the monopolist. If equity matters, regulators may oppose such practices.
• •Third-degree PD can increase or decrease total welfare relative to single-price monopoly — compute outputs to decide.
• •Two-part tariffs can implement efficiency at the margin ($p=MC$) while allowing complete extraction through the lump-sum fee when consumers are identical; with heterogeneity, selective exclusion may occur.

Downstream connections (what this enables you to analyze next)

Mastering price discrimination is essential for advanced topics in industrial organization and mechanism design. Specific forward-looking areas that build directly on this lesson:

• •Mechanism design and optimal auctions: understanding how to design menus under IC/IR is central to optimal auction theory and optimal monopoly regulation.
• •Dynamic pricing and personalized pricing with learning: first-degree PD approximations emerge when firms can learn individual demand functions from data.
• •Ramsey pricing and regulated monopolies: optimizing welfare subject to revenue constraints (regulation) uses the elasticity–markup tradeoffs seen in third-degree PD.
• •Antitrust analysis: merger and exclusion effects sometimes hinge on the ability to price discriminate across products and segments.

Link back to prerequisites

This lesson builds on "In Profit Maximization, we learned" the necessary MR=MC first-order conditions and constrained optimization logic for IC and IR constraints. It also uses "In Consumer Surplus, we learned" how to compute and interpret areas under demand curves to evaluate welfare changes, consumer surplus, and deadweight loss. When doing numerical exercises below, explicitly solve MR=MC and compute CS areas as you learned in those prerequisites.