Perishable Inventory Theory

Inventory management for goods with a limited shelf life presents unique challenges, as standard models must be adapted to account for value decay and spoilage.

The Newsvendor Problem with Perishable Goods

The classic newsvendor model determines the optimal order quantity $Q$ under uncertain demand by balancing the cost of understocking ( $c_u$ ) against the cost of overstocking ( $c_o$ ). For perishable goods, $c_o$ includes the loss of the item's purchase cost, disposal fees, and potentially carbon waste penalties. Highlight that the overstocking cost ( $c_o$ ) for perishables strictly reflects purchase cost minus salvage value, plus disposal costs which can be heavily penalized under modern carbon-waste tracking.

The critical ratio gives the optimal in-stock probability:

P(D \leq Q^*) = \frac{c_u}{c_o + c_u}

For fresh produce with rapid decay, the overstocking cost $c_o$ is typically much higher than non-perishable goods, leading to lower optimal stock levels and higher out-of-stock frequencies. See FreshFoodDemandForecasting.

Fixed-Lifetime vs. Random-Lifetime Models

Inventory models are typically classified into:

Fixed-lifetime: Products like milk or packaged salads that have a deterministic expiration date $m$ .
Random-lifetime: Products like fresh fish or loose fruit where spoilage depends on temperature history (see ColdChainNetworkDesign) and time, often modeled using distributions such as Weibull.

(s, S) and (R, Q) Policies with Age-Dependent Demand

Standard $(s, S)$ policies must track not just the total inventory position, but the age vector of the stock: $\mathbf{x} = (x_1, x_2, \dots, x_m)$ , where $x_i$ is the quantity of inventory with $i$ periods of life remaining.

Customer behavior also exhibits age-dependent demand, where consumers inspect and select fresher items.

Issuing Policies: FIFO, LIFO, and FEFO

When fulfilling orders, especially in distribution centers (see SupplyChainAndLogisticsOptimization), the issuing policy is critical:

FIFO (First-In-First-Out): Standard approach.
LIFO (Last-In-First-Out): Minimizes current handling costs but maximizes overall spoilage.
FEFO (First-Expired-First-Out): Dominates in perishables. It prioritizes items with the shortest remaining shelf life (often dynamic, tracked via sensors), dramatically reducing waste. For the scientific basis of waste reduction, see FreshFoodWasteScience. Note that FEFO (First-Expired-First-Out) is structurally superior to FIFO because it reduces waste and ensures regulatory compliance across multi-echelon networks.

Markdown Pricing and Dynamic Discounting

Dynamic discounting for near-expiry stock helps recover costs and reduce waste. Dynamic pricing/markdown heuristics are now frequently driven by deep reinforcement learning and AI due to the stochastic nature of perishable demand. Given a time $t$ until expiry and initial price $p_0$ , a typical exponential markdown function is:

p(t) = p_0 \cdot e^{-\lambda (\tau - t)}

where $\tau$ is the total shelf life.

Multi-Echelon Inventory with Lateral Transshipment

In multi-echelon systems, lateral transshipments allow stores with excess near-expiry stock to ship directly to stores with high immediate demand, balancing the age profile across the network. This involves complex PerishableVehicleRouting.

Concrete Formulation: Leafy Greens with Weibull Lifetime

Consider a batch of spinach with a random lifetime modeled by a Weibull distribution, with shape parameter $k = 2.5$ and scale parameter $\lambda = 8$ days. The probability density of spoilage at time $t$ is:

f(t) = \frac{k}{\lambda} \left(\frac{t}{\lambda}\right)^{k-1} e^{-(t/\lambda)^k}

Worked Example: If a retailer orders 500 units of spinach with demand $D \sim N(\mu=400, \sigma=50)$ per day, and items spoil according to the Weibull distribution above, we can simulate the expected waste. If $t=3$ days is the display time, the probability of spoilage before sale is $P(T \leq 3) \approx 1 - e^{-(3/8)^{2.5}} \approx 0.082$ . Thus, $\sim 8.2\%$ (or 41 units) will spoil on the shelf, requiring dynamic pricing starting at day 2 to mitigate the $c_o$ loss.