Elsevier

Operations Research Letters

Volume 38, Issue 5, September 2010, Pages 436-440
Operations Research Letters

A note on demand functions with uncertainty

https://doi.org/10.1016/j.orl.2010.06.001Get rights and content

Abstract

Demand functions with “additive” and “multiplicative” uncertainty can fail to satisfy the “Slutsky symmetry” conditions required for demand to be based on utility maximization. This makes welfare analysis infeasible, which is important in many applications. Additive uncertainty can also generate negative demand realizations, and this can create non-trivial analytical problems.

Introduction

As Kreps [15] notes, “the central figure in microeconomic theory is the consumer”, defined as “an entity who chooses from some given set of feasible options”. Economic analysis, therefore, begins with a model of consumer behavior. The theory of demand is developed from these models, and demand functions used in applied economic analysis are based on an underlying model of consumer behavior.

In many applications, however, functional forms for demand are specified without making the underlying consumer behavior model explicit. A particular demand function is guaranteed to be based on a utility-maximizing representative consumer if and only if certain conditions on the partial derivatives of demand with respect to prices are satisfied. These conditions, known as Slutsky symmetry, are described in Section 2.

A key aspect of demand in operations management models, as opposed to traditional models in economics, is the presence of uncertainty. Random-demand functions are sometimes generated from an underlying random-utility consumer model, for example from an “address” (Hotelling-type) model. Most papers in the operations literature, however, simply start with a particular functional form for random demand.

Consider the demand for a product i. In its most general form, the demand function with uncertainty can be represented as Di(p,ϵi), where p is the price vector and ϵi is the uncertainty parameter. In most cases, however, uncertainty is introduced into deterministic demand functions through an “additive” or “multiplicative” uncertainty parameter (see [24]), or a combination of both (e.g. [12], [7], [1]). In the additive uncertainty model, Di(p,ϵi)=Di(p)+ϵi; in the multiplicative model, Di(p,ϵi)=Di(p)ϵi. The underlying, deterministic, demand functions, Di(p), are usually well-established demand functions from the economics literature.

Demand functions with additive or multiplicative uncertainty do not, in general, satisfy the conditions necessary to guarantee utility maximization by a representative consumer. This issue has largely been overlooked in the literature. Two specific technical problems arise.

First, these demand functions may not satisfy the “Slutsky symmetry” conditions. While it is not necessary for aggregate or market demands to satisfy Slutsky symmetry, these conditions must be satisfied if we assume that demand functions are generated by a “representative consumer” (see [21]). These conditions are routinely imposed on demand functions in applied economics models because, without a clear understanding of the underlying utility model, which the representative consumer model provides, it is impossible to do a meaningful analysis of consumer surplus and total welfare. The inability to perform welfare analysis is a significant handicap for operations researchers working in areas where welfare implications are important, such as in contract theory and antitrust regulation and in other public sector applications.

Second, models with additive uncertainty allow negative demand to occur in some states of the world. Unfortunately, the imposition of non-negativity constraints is not trivial. Non-negativity constraints can create problems with the sufficiency of first-order conditions in monopoly models, and the existence of equilibria in an oligopoly. These challenges have been ignored in some operations models. The aim of this methodological note is to set out more carefully the constraints that must be met in operations models that adopt demand functions as their starting point.

The paper proceeds as follows. After presenting a brief overview of demand theory, I discuss (1) the potential failure of Slutsky symmetry in commonly used demand functions with uncertainty and (2) the problems associated with the imposition of non-negativity constraints in additive demand models.

Section snippets

Overview of classical demand theory

This section presents a brief overview of classical demand theory. For more details, the reader is referred to the microeconomic texts by [15], [21], [28], from which this section draws.

Failure of symmetry in demand functions with uncertainty

In a model with demand uncertainty, note that symmetry has to be satisfied for all realizations of demand. For example, suppose that the demand for product 1 and product 2 is D1(p1,p2,ϵ1) and D2(p1,p2,ϵ2), respectively. The symmetry condition requires that D1(p1,p2,ϵ1)/p2=D2(p1,p2,ϵ2)/p1 for every pair of realized random variables (ϵ1,ϵ2).

In the multiplicative case, (D1(p1,p2)/p2)ϵ1(D2(p1,p2)/p1)ϵ2 even if (D1(p1,p2)/p2)=(D2(p1,p2)/p1) unless ϵ1=ϵ2. In other words, even if the

Imposition of a non-negativity constraint on demand

Consider a single-product monopolist (the price-setting newsvendor) who has to choose price and inventory before observing demand. Assume that the firm faces the demand function D(p,ϵ), which is non-increasing in p. Define the “choke” price, pchoke, as the price at which demand falls to zero. (Karlin and Carr [14] refer to this as the “null” price.) In traditional demand models, the choke price can be finite (e.g. linear demand; D(p)=abp) or infinite (e.g. iso-elastic demand; D(p)=apb). Note

Discussion and conclusions

For demand functions with additive and multiplicative uncertainty, the failure of Slutsky symmetry makes welfare analysis infeasible. Many models in the supply chain contracting literature, however, rely on these functions. Research based on these models, unfortunately, will be unable to influence public policy discussions about the welfare implications of contracts and will therefore be unable to impact contracting practice. In addition, models with additive uncertainty often ignore the

Acknowledgement

I thank the Associate Editor, Maurice Queyranne and Ralph Winter for several helpful comments and suggestions.

References (29)

  • V. Agrawal et al.

    Impact of uncertainty and risk aversion on price and order quantity in the newsvendor problem

    Manufacturing & Service Operations Management

    (2000)
  • K. Anand et al.

    The strategic perils of delayed differentiation

    Management Science

    (2007)
  • F. Bernstein et al.

    Decentralized supply chains with competing retailers under demand uncertainty

    Management Science

    (2005)
  • G.P. Cachon et al.

    Purchasing, pricing, and quick response in the presence of strategic consumers

    Management Science

    (2008)
  • D.W Carlton

    Market behavior with demand uncertainty and price inflexibility

    American Economic Review

    (1978)
  • Hung-Po Chao

    Exhaustible resource models: the value of information

    Operations Research

    (1981)
  • X. Chen et al.

    Coordinating inventory control and pricing strategies with random demand and fixed ordering cost: the finite horizon case

    Operations Research

    (2004)
  • J. Chod et al.

    Resource flexibility with responsive pricing

    Operations Research

    (2005)
  • A. Ehrenmann et al.

    A comparison of electricity market designs in networks

    Operations Research

    (2009)
  • A. Farahat et al.

    Price competition among multiproduct firms

  • Federal Trade Commission (2008), Guide to The Antitrust Laws, Retrieved October 22, 2009, from...
  • A. Federgruen et al.

    Combined pricing and inventory control under uncertainty

    Operations Research

    (1999)
  • J.R. Hicks

    The rehabilitation of consumers’ surplus

    Review of Economic Studies

    (1941)
  • S. Karlin et al.

    Prices and optimal inventory policy

  • Cited by (13)

    • The ubiquitous nature of inventory: Vendor Managed Consignment Inventory in adverse market conditions

      2021, European Journal of Operational Research
      Citation Excerpt :

      VMCI with a non–negativity constraint becomes non–trivial and less tractable than VMCI without this assumption and more complicated than the classical newsvendor model with a similar constraint. Under the non–negativity constraint, there is no guarantee that the retailer’s expected profit is quasi–concave with respect to the retail price (Krishnan, 2010). The papers that do not imply non–negativity and allow for the possibility of negative demand, do not have this technical problem.

    • The price-setting newsvendor problem with nonnegative linear additive demand

      2018, European Journal of Operational Research
      Citation Excerpt :

      Xu, Chen, and Xu (2010) analyzed the effect of demand uncertainty in a price-setting newsvendor model and Zhao and Atkins (2008) studied an equilibrium extension of the newsvendor problem with multiple competitive newsvendors. Krishnan (2010) pointed out that a nonnegativity constraint should be imposed on the total demand in the linear additive demand formulation in Petruzzi and Dada (1999) to eliminate the possibility of negative actual demand. Otherwise, the computed optimal price p may be suboptimal and the expected profit is underestimated because the actual demand is “choked off” at high optimal p values.

    • The Pandemic-Type Demand Shocks in the Mean-Variance Newsvendor Problem

      2022, Management and Production Engineering Review
    View all citing articles on Scopus
    View full text