Hotelling Spatial Competition — Interactive Pricing Theory

When consumers must travel to reach a seller, firms gain local market power even when their products are otherwise identical. This insight, formalized by Hotelling (1929), is the foundation of spatial competition theory and, more broadly, of models of horizontal product differentiation. The same mathematics that describes two ice-cream vendors on a beach also explains why software products cluster around similar feature sets, why political candidates converge to the median voter, and how transport costs translate into pricing power.

Introduction

Consider two competing firms selling an identical good. In the standard Bertrand model, they undercut each other until price equals marginal cost. Yet in practice, firms facing direct competition still earn positive margins. One explanation is differentiation: if consumers care about more than price—location, style, feature set, or any other product attribute—each firm acquires a captive segment that is reluctant to switch.

Hotelling (1929) captured this idea elegantly. Place consumers uniformly along a line segment representing a spectrum of preferences (or a literal street), and put two firms somewhere on that line. Each consumer buys from the nearer firm, paying both the posted price and a “transport cost” proportional to distance. Because switching is costly, each firm has local monopoly power over the consumers nearest to it.

This topic develops the linear city model in full, examines the famous minimum differentiation claim and why it fails, derives the circular city equilibrium of Phillips (2021), and surveys the applications that have made spatial competition the workhorse framework for horizontal differentiation.

For zone-based surge pricing applications, see Location-Based Pricing. For the congestion externality and Pigouvian tolls, see Congestion & Pigouvian Pricing.

The Linear City Model

Hotelling’s (1929) model places consumers uniformly along a line segment $[0, 1]$ and two firms at locations $a$ and $b$ with $0 \le a < b \le 1$ . Each consumer purchases from the firm that minimizes their total cost: the price charged plus a transport cost proportional to distance.

Definition — Linear City Model

A spatial competition model in which consumers are uniformly distributed on

[0,1]

and two firms at locations

a

and

b

set prices

p_1

and

p_2

. A consumer at position

x

chooses firm

i

to minimize

p_i + t \cdot |x - \ell_i|

, where

t > 0

is the transport cost per unit distance and

\ell_i

is the firm’s location.

The indifferent consumer—the one who is equally well off buying from either firm—is located at:

x^* = \frac{p_2 - p_1 + t(a + b)}{2t}

(1)

Firm 1 captures all consumers in $[0, x^*]$ and firm 2 captures $[x^*, 1]$ . The firms’ demands are therefore $D_1 = x^*$ and $D_2 = 1 - x^*$ . Notice that raising $p_1$ shifts $x^*$ leftward, shrinking firm 1’s market share continuously rather than dropping it to zero (as in standard Bertrand competition). This smoothness is what allows a Nash equilibrium in prices to exist.

Nash Equilibrium in Prices

Each firm maximizes profit $\pi_i = (p_i - c_i) \cdot D_i$ taking the rival’s price as given. Solving the first-order conditions simultaneously yields the Nash equilibrium prices:

p_1^* = t(b - a) + \frac{2c_1 + c_2}{3}, \qquad p_2^* = t(b - a) + \frac{c_1 + 2c_2}{3}

(2)

Hotelling Equilibrium Prices

In the linear city with firms at locations

a

and

b

(

a < b

) and marginal costs

c_1, c_2

, the unique Nash equilibrium in prices satisfies

p_i^* = t(b-a) + (2c_i + c_j)/3

. When firms are symmetric (

c_1 = c_2 = c

) and located at the endpoints (

a = 0, b = 1

), this simplifies to

p^* = c + t

, and each firm serves exactly half the market.

Several properties of this equilibrium merit attention. First, prices increase in the transport cost $t$ : when consumers find it costly to travel, firms enjoy greater local market power. Second, prices increase in the distance $b - a$ between firms: more differentiated firms face less intense competition and therefore charge higher prices. Third, each firm’s equilibrium price loads more heavily on its own cost than on the rival’s cost—a one-dollar increase in $c_1$ raises $p_1^*$ by $\tfrac{2}{3}$ but raises $p_2^*$ by only $\tfrac{1}{3}$ . Cost passthrough is incomplete in both directions.

Interactive: Linear City

The visualization below shows the total cost faced by a consumer at each location along the city. The V-shaped curves represent $p_i + t \cdot |x - \ell_i|$ for each firm. The consumer at the intersection of the two curves is the indifferent consumer $x^*$ . Below the chart, a color-coded bar shows which firm captures each market segment.

Increase the transport cost $t$ to see how spatial differentiation raises equilibrium prices. Move the firms closer together to observe intensified competition and lower margins. Asymmetric costs shift the indifferent consumer toward the higher-cost firm.

The Beach Vendor Problem

Two ice cream vendors on a beach of unit length, with zero marginal cost and transport cost

t = 5

, each set a Nash price of

p^* = 0 + 5 = \$5

and share the market equally. If vendor 1 moves from position 0 to position 0.25, the firms are closer together (

b - a = 0.75

), so equilibrium prices fall to

p^* = 5 \times 0.75 = \$3.75

. Consumers benefit from less differentiation through lower prices, but those near the ends of the beach face higher transport costs than before.

The Minimum Differentiation Paradox

Hotelling’s original paper made an influential claim: when firms can choose their locations, both will converge to the center of the city (positions $a = b = \tfrac{1}{2}$ ). This “principle of minimum differentiation” seemed to explain why competing stores, political parties, and products tend to cluster around the median consumer.

The intuition is straightforward. If firm 1 is to the left of center and firm 2 is at the center, firm 1 can steal customers by moving slightly rightward toward the center, capturing consumers between its old and new positions without losing any consumers to the left (there are no competitors on that side). The same logic applies symmetrically to firm 2. The only stable position is both firms at the center.

However, d’Aspremont, Gabszewicz, and Thisse (1979) demonstrated that this reasoning breaks down when firms are close together. The problem is price undercutting: when $a$ and $b$ are nearly equal, there is almost no spatial buffer between the firms, so either firm can capture the entire market by cutting price by a small amount. This triggers Bertrand-style price competition that eliminates profits, making the central location unattractive. The minimum differentiation equilibrium does not survive the joint optimization over prices and locations.

Quadratic Transport Costs

d’Aspremont, Gabszewicz, and Thisse (1979) showed that the existence of a price equilibrium—and the direction of location equilibrium—depend critically on the shape of the transport cost function. Under linear transport costs $t \cdot |x - \ell_i|$ , no Nash equilibrium in prices exists when the firms are sufficiently close, because the best response functions fail to intersect (demand jumps discontinuously as one firm undercuts the other).

Under quadratic transport costs $t \cdot (x - \ell_i)^2$ , demand is always smooth and a price equilibrium exists for all firm locations. The indifferent consumer under quadratic costs is:

x^* = \frac{1}{2} + \frac{p_2 - p_1}{2t(b - a)}

(3)

With quadratic costs, the location game has the opposite conclusion: the subgame-perfect equilibrium is maximum differentiation, with firms locating at the two endpoints ( $a = 0$ , $b = 1$ ). The logic reverses: moving away from the center softens price competition by increasing the effective switching cost for consumers near the boundary between firms.

Definition — Minimum vs. Maximum Differentiation

Under linear transport costs, no price equilibrium exists near the center, so the principle of minimum differentiation fails. Under quadratic transport costs, equilibrium exists everywhere, and the location equilibrium features maximum differentiation (firms at the endpoints). The key takeaway is that differentiation is a strategic choice that trades off market share against the intensity of price competition.

This result has broad implications. In product markets, it predicts that competing firms will voluntarily differentiate their offerings to relax price competition—a phenomenon visible in premium-versus-value brand positioning, geographic market segmentation, and feature differentiation in software. The transport cost parameter $t$ captures how much consumers value variety: high $t$ means consumers strongly prefer their ideal product, giving firms more pricing power and stronger incentives to differentiate.

The Circular City (Salop Model)

Salop (1979) extended Hotelling’s model to a circle of unit circumference with $n$ firms distributed equidistantly. The circular geometry eliminates the asymmetry between endpoint and interior firms and is the natural model for markets where there is no inherent ordering of consumer preferences.

With $n$ symmetric firms, quadratic transport cost $t$ , and free entry that drives profit to zero, the equilibrium price is:

p^* = c + \frac{t}{n}

(4)

Each firm serves a market segment of length $1/n$ , and the price-cost margin equals $t/n$ . Three results follow immediately:

Free entry: As the number of firms $n$ increases, $p^* \to c$ . In the limit, price competition drives margins to zero, echoing the Bertrand result for undifferentiated goods.
Higher transport cost: Greater $t$ means consumers value product matching more, so each firm commands a larger premium for being the closest option.
Social optimum vs. free entry: Free entry may lead to too many or too few firms depending on whether the business-stealing externality (firms take customers from rivals) dominates the variety externality (each new firm improves the match quality for some consumers). The Salop model is a standard framework for analyzing these offsetting forces in industrial organization.

Salop Circular City Equilibrium

With

n

symmetric firms equidistant on a unit circle, transport cost

t

, and marginal cost

c

, the symmetric Nash equilibrium price is

p^* = c + t/n

. The equilibrium number of firms under free entry satisfies the zero-profit condition, yielding

n^* = \sqrt{t / F}

where

F

is the fixed cost of entry.

The Salop model is the preferred framework when:

There is no natural “endpoint” in the preference space (brands wrap around, as with product varieties on a color wheel or political positions in a circular spectrum).
Entry and exit are important: the zero-profit condition pins down the equilibrium number of competitors.
The analyst wants a clean closed-form relationship between the number of competitors, transport costs, and the price-cost margin.

Applications

The Hotelling framework has been applied in settings far removed from literal geography:

Retail location. Supermarkets, pharmacies, and fast-food chains choose locations partly to avoid head-to-head competition. The transport cost represents the consumer’s disutility from traveling to a non-preferred location.
Product differentiation. Competing software products position their feature sets along a quality or style dimension. The “transport cost” is the hassle of using a product that does not match the user’s preferred workflow. High transport costs explain why competing productivity applications can coexist at premium prices despite being functionally similar.
Media and political positioning. The original Hotelling article used political parties as an example. Two parties competing for median-voter support tend to converge toward the center of the political spectrum—the linear city analogy is direct.
Airline routes and hotel chains. Airlines compete on schedule convenience (departure times as locations on a time line) and hotel chains compete on neighborhood coverage. Both industries show the dynamics of differentiation and spatial market power predicted by the model.

Smartphone Ecosystem Pricing

Consider two smartphone ecosystems positioned at opposite ends of a preference spectrum (say, openness vs. integration). With a consumer population distributed uniformly between these extremes and a high switching cost (transport cost

t

representing the cost of migrating apps, contacts, and habits), the Hotelling equilibrium predicts premium prices and stable market shares even when the underlying hardware components are nearly identical. Reducing the switching cost (e.g., through cross-platform app availability) shifts the model toward Bertrand competition, compressing margins.

References

d’Aspremont, C., Gabszewicz, J. J. & Thisse, J.-F. (1979). “On Hotelling’s “Stability in Competition”.” Econometrica, 47(5), 1145–1150.
Hotelling, H. (1929). “Stability in Competition.” Economic Journal, 39(153), 41–57.
Phillips, R. L. (2021). Pricing and Revenue Optimization, 2nd ed.. Stanford University Press.

d’Aspremont, C., Gabszewicz, J. J. & Thisse, J.-F. (1979). On Hotelling’s “Stability in Competition.” Econometrica, 47(5), 1145–1150.
Salop, S. C. (1979). Monopolistic Competition with Outside Goods. Bell Journal of Economics, 10(1), 141–156.