Pricing Under Demand Uncertainty

Most pricing models assume the firm knows how many units it will sell at each price. In practice, a fashion retailer ordering for next season, a newspaper deciding tomorrow’s print run, or a bakery choosing how many croissants to bake all face the same problem: they must commit to both a price and an inventory quantity before demand is realized.

The Problem

The price-setting newsvendor problem extends the classical newsvendor model by making price an endogenous decision variable. Instead of taking the selling price as given and choosing only how many units to stock, the firm simultaneously selects a price $p$ and an order quantity $Q$ to maximize expected profit under uncertain demand.

This joint optimization was first studied by Whitin (1955) and has since become a foundational model in operations-pricing integration. The key insight is that demand uncertainty fundamentally changes the optimal pricing decision: the stochastic optimal price differs from the price that would be optimal under certainty. As Petruzzi and Dada (1999) showed in their comprehensive review, the interaction between price and quantity decisions creates effects that neither decision captures alone.

The Classical Newsvendor

Before introducing the price decision, consider the classical setting. A firm sells a perishable product at a fixed price $p$ , pays a unit cost $c$ for each unit ordered, and recovers a salvage value $v < c$ for each unit left unsold. Demand $D$ is a random variable with known distribution $F$ .

Definition — Critical Fractile

The critical fractile (or critical ratio) is the ratio of the underage cost to the sum of underage and overage costs:

\mathrm{CF} = \frac{p - c}{p - v}

where $p - c$ is the marginal profit from selling one more unit (cost of understocking) and $c - v$ is the marginal loss from having one unsold unit (cost of overstocking). The ratio balances the risk of ordering too few against the risk of ordering too many.

The optimal order quantity $Q^*$ satisfies the condition that the probability of meeting all demand equals the critical fractile:

Q^* = F^{-1}\!\left(\frac{p - c}{p - v}\right)

(1)

When demand is normally distributed with mean $\mu$ and standard deviation $\sigma$ , this becomes $Q^* = \mu + \sigma \, \Phi^{-1}(\mathrm{CF})$ where $\Phi^{-1}$ is the standard normal quantile function. Note that when margins are high ( $\mathrm{CF}$ close to 1), the firm orders well above the mean; when margins are thin, it orders conservatively below the mean.

Adding the Price Decision

Now suppose the firm can also choose the selling price. Demand depends on price through a deterministic function scaled by a random shock. We model demand as:

D(p) = d(p) + \varepsilon

(2)

where $d(p)$ is the expected demand at price $p$ (for instance, a constant-elasticity function $d(p) = a(p_0 / p)^\varepsilon$ ) and $\varepsilon$ is a mean-zero random variable capturing demand uncertainty. The firm’s expected profit given a price $p$ and order quantity $Q$ is:

\Pi(p, Q) = \mathbb{E}\bigl[(p - v)\min\bigl(D(p),\, Q\bigr) - (c - v)\,Q\bigr]

(3)

The first term captures revenue net of salvage on units actually sold (constrained by inventory), while the second term is the net ordering cost. The firm seeks:

\max_{p,\, Q} \;\; \Pi(p, Q) = \mathbb{E}\bigl[(p - v)\min\bigl(D(p),\, Q\bigr) - (c - v)\,Q\bigr]

(4)

This problem cannot generally be decomposed into separate price and quantity decisions. The optimal price depends on the stocking level (because a higher price reduces the chance of a stockout, lowering the effective cost of understocking), and the optimal quantity depends on the price (because the critical fractile changes with $p$ ). As Federgruen and Heching (1999) established, the two decisions must be solved jointly.

Joint Optimization Results

Uncertainty Raises the Optimal Price

Under additive or multiplicative demand uncertainty with a constant-elasticity price-response function, the optimal stochastic price $p^*_S$ exceeds the deterministic optimal price $p^*_D$ :

p^*_S \;\ge\; p^*_D = \frac{\varepsilon}{\varepsilon - 1}\, c

where $\varepsilon$ is the price elasticity of demand. The deterministic optimum is the familiar markup rule from monopoly pricing; the stochastic optimum adds an upward correction to hedge against demand uncertainty.

The intuition is that uncertainty introduces an asymmetric cost structure. When the firm orders too many units, it loses $c - v$ per unsold unit. When it orders too few, it loses $p - c$ per unit of unmet demand. By raising the price, the firm reduces expected demand (and hence the optimal order quantity), which lowers overage risk. The higher margin per unit also makes each sale more valuable, partially compensating for lower volume.

The combined effect is that the firm hedges against uncertainty by simultaneously raising price and reducing quantity relative to the deterministic solution. This hedging behavior is especially pronounced when overage costs are high relative to underage costs, as analyzed by Kocabıyıkoğlu and Popescu (2011).

Fashion Retailer

Consider a fashion retailer sourcing winter coats at a unit cost of $c = \$10$ with a salvage value of $v = \$2$ (end-of-season clearance). Expected demand at the reference price of $30 is 200 units, with a standard deviation of 40 units and a price elasticity of 2.5. Under these parameters:

Deterministic optimum: The monopoly markup rule gives $p^*_D = \frac{2.5}{1.5} \times 10 = \$16.67$ with quantity equal to mean demand at that price.
Stochastic optimum: The joint optimizer finds a higher price (roughly $18–$20) and a lower quantity, reflecting the cost of unsold inventory at end of season.
Profit gap: Expected profit under uncertainty is lower than the deterministic profit, with the gap representing the cost of demand uncertainty.

Try the interactive optimizer below to see the exact values and how they shift with parameters.

Interactive Joint Optimizer

The chart below sweeps across prices, computing the expected profit under both the stochastic model (solid line) and the deterministic model (dashed line) at each price point. The stochastic profit at each price uses the optimal newsvendor quantity for that price. Dots mark the respective optimal prices. Adjust cost, salvage value, demand variability, and price elasticity to observe how the profit gap and price differential respond.

The Effect of Uncertainty

Monotonicity in Demand Variance

As the standard deviation of demand $\sigma$ increases (holding the mean fixed), the optimal stochastic price $p^*_S(\sigma)$ is nondecreasing and the optimal order quantity $Q^*_S(\sigma)$ is nonincreasing:

\frac{\partial p^*_S}{\partial \sigma} \ge 0 \qquad\text{and}\qquad \frac{\partial Q^*_S}{\partial \sigma} \le 0

In the limit as $\sigma \to 0$ , the stochastic solution converges to the deterministic solution. As $\sigma \to \infty$ , the firm orders near zero and charges a very high price to extract maximum value from each rare sale.

This monotonicity result, which follows from the structure of the newsvendor expected-profit function, has a natural economic interpretation. Greater uncertainty increases both the expected overage (unsold inventory) and the expected underage (lost sales). The firm responds by shifting its strategy toward higher margins and lower volume. Each unit sold is more profitable, but fewer units are stocked, reducing the downside from excess inventory.

The two-panel chart below traces the optimal price and quantity as $\sigma$ increases from near zero (essentially deterministic) to very high uncertainty. The horizontal dashed lines show the deterministic benchmarks. Observe that the divergence from the deterministic solution accelerates as $\sigma$ grows: the cost of ignoring uncertainty rises nonlinearly.

Interactive Uncertainty Explorer

The two-panel chart sweeps demand standard deviation from 5 to 120, computing the jointly optimal price and quantity at each level. The top panel shows optimal price; the bottom panel shows optimal quantity. Horizontal dashed lines mark the deterministic benchmarks. Adjust cost and elasticity to see how they modulate the sensitivity to uncertainty.

Practical Implications

Fashion Retail

Fashion retailers face some of the highest demand uncertainty in commerce: trends shift quickly, lead times are long, and unsold inventory must be cleared at deep discounts. The newsvendor model explains why fashion brands charge high initial markups. A typical fashion item with a 60% end-of-season markdown and high demand variance will have a stochastic optimal price 10–20% above the deterministic optimum. This margin buffer absorbs the cost of unsold inventory.

Seasonal and Perishable Products

Products with a single selling season and low salvage value—holiday decorations, event merchandise, fresh produce—exhibit the strongest newsvendor effects. The salvage value is often near zero (expired food, outdated seasonal goods), which pushes the overage cost close to the full unit cost $c$ . The interactive optimizer shows that as salvage drops toward zero, the gap between stochastic and deterministic prices widens substantially.

Digital and Print Media

Newspapers and magazines face the original newsvendor problem: each day’s print run must be committed before sales are observed. Unsold copies are recycled at negligible salvage value. Even though digital distribution has reduced the relevance of physical inventory, the model applies to any setting where capacity or inventory must be committed before demand is known—including cloud computing resource provisioning and event ticket pricing.

From Theory to Practice

The central lesson of the price-setting newsvendor is that ignoring demand uncertainty leads to prices that are too low and order quantities that are too high. Firms that set prices using deterministic models and then separately determine order quantities are implicitly underpricing their products. As Phillips (2021) emphasizes, integrating the pricing and inventory decisions is essential whenever demand uncertainty is significant and excess inventory is costly.

References

Federgruen, A. & Heching, A. (1999). “Combined Pricing and Inventory Control Under Uncertainty.” Operations Research, 47(3), 454–475.
Kocabıyıkoğlu, A. & Popescu, I. (2011). “An Elasticity Approach to the Newsvendor with Price-Sensitive Demand.” Operations Research, 59(2), 301–312.
Petruzzi, N. C. & Dada, M. (1999). “Pricing and the Newsvendor Problem: A Review with Extensions.” Operations Research, 47(2), 183–194.
Phillips, R. L. (2021). Pricing and Revenue Optimization, 2nd ed.. Stanford University Press.
Talluri, K. T. & van Ryzin, G. J. (2004). The Theory and Practice of Revenue Management. Springer.
Whitin, T. M. (1955). “Inventory Control and Price Theory.” Management Science, 2(1), 61–68.