Abstract
In this paper, we study distributions that describe markets with linear stochastic demand. We express the price elasticity of expected demand in terms of the mean residual demand (MRD) function of the demand distribution and characterize optimal prices or equivalently, points of unitary elasticity, as fixed points of the MRD function. This leads to economic interpretable conditions on the demand distribution under which such fixed points exists and are unique. In particular, markets with increasing price elasticity of expected demand that eventually become elastic correspond to distributions with decreasing generalized mean residual demand (DGMRD) and finite second moment. DGMRD distributions strictly generalize the widely used increasing generalized failure rate (IGFR) distributions. In real life economic applications, they arise naturally as mixtures of (possibly) IGFR distributions over disjoint intervals. We further elaborate on the relationship of the two classes and link their limiting behavior at infinity. We examine moment and closure properties of the DGMRD distributions that are important in economic applications and illustrate our results with examples.
Introduction
In revenue management and economics, identifying the "perfect price" is a fundamental challenge, particularly under demand uncertainty. For a monopolistic seller, a primary objective is to ensure that the revenue function is unimodal—meaning it has exactly one peak. Unimodality guarantees that a unique optimal price exists and prevents market instability where minor shifts in conditions could lead to drastic changes in the optimal pricing strategy.
Historically, researchers have relied on the Increasing Generalized Failure Rate (IGFR) property to guarantee this single peak. However, IGFR is restrictive; it requires the demand distribution to be defined over a single, connected interval. This limitation makes it unsuitable for modeling "gappy" demand scenarios, such as markets composed of distinct consumer segments (e.g., budget vs. luxury) separated by a valuation gap. This paper introduces the Decreasing Generalized Mean Residual Demand (DGMRD) class, a more flexible framework that accommodates such complexities while maintaining the guarantee of a unique optimal price.
The DGMRD Framework
The authors analyze markets characterized by linear stochastic demand, where expected revenue depends on price $p$ and a random demand level $\alpha$. A central component of this analysis is the Mean Residual Demand (MRD) function, $m(p)$, which represents the expected additional demand from consumers whose valuations exceed $p$.
The core innovation is the Generalized Mean Residual Demand (GMRD) function, defined as $\ell(p) = m(p)/p$. A distribution is classified as DGMRD if $\ell(p)$ is decreasing. The authors prove that for any DGMRD distribution with a finite second moment ($E\alpha^2 < \infty$), the revenue function is guaranteed to be unimodal.
The Fixed Point Rule
A significant result of this framework is the characterization of the optimal price $p^*$. The authors demonstrate that $p^*$ is the fixed point of the MRD function, where $p = m(p)$. This point corresponds to unitary price elasticity, providing a clear economic interpretation for the optimal pricing condition.
Hierarchy and Generalization
The DGMRD class is a strict generalization of the IGFR class. While every IGFR distribution is DGMRD, the reverse is not true. This relationship is part of a broader hierarchy of distribution classes used in reliability and economic theory.
As shown in Figure 1, the DGMRD class sits at the end of a chain of implications starting from the more restrictive Increasing Failure Rate (IFR) and Decreasing Mean Residual Demand (DMRD) classes. This hierarchy allows researchers to apply the DGMRD framework to a wider variety of realistic market models that were previously mathematically intractable.
Illustrative Examples and Robustness
To demonstrate the broader reach of DGMRD, the authors examine distributions that fail the IGFR condition but satisfy the DGMRD property. A prime example is the Birnbaum-Saunders distribution.
In Figure 2, the left plot illustrates a non-monotonic generalized failure rate, disqualifying the distribution from the IGFR class. However, the right plot shows the monotonic behavior of price elasticity, which corresponds to the DGMRD property, proving that this class can handle demand patterns that IGFR cannot.
Furthermore, the DGMRD class is robust under mixtures. The authors show that mixing different demand scenarios—even those with gaps in consumer valuations—can result in a DGMRD distribution. This allows for sophisticated scenario analysis in revenue management that was previously difficult to model.
Conclusion
The introduction of the DGMRD class provides a significant advancement in the mathematical modeling of demand. By requiring only a finite second moment and a decreasing GMRD function, economists can model complex, multi-segment markets with the assurance of a unique optimal price. While currently focused on linear demand, this framework offers a more resilient toolkit for modern revenue management.
Summary updated