Andy A. Tsay

We live in an uncertain world, and nearly every rational behavior includes some acknowledgment that things might not turn out as expected. Some actions, such as the purchase of insurance policies or financial hedging, would cease to exist altogether in a risk-free world. An understanding of how decisions can be colored by sensitivity to risk is therefore imperative to any student of commerce. This is of particular relevance with regards to the structuring of relationships between partners. Different parties might have different sensitivities to risk due to financial strength, depth/breadth of assets and activities, access to hedging instruments, etc.

An interesting case study is the vertical supply/distribution agreements between independent firms (whose parties are referred to as a manufacturer and a retailer). In this context, manufacturer return policies have been identified by both the academic and practitioner literatures as a popular vehicle for addressing the risk engendered by market demand uncertainty in a variety of industries (e.g., books, magazines, newspapers, recorded music, computer hardware and software, greeting cards, and pharmaceuticals). The design of such a policy must consider a number of questions. How is each party's behavior colored by sensitivity to risk? When is each party better off facing the demand risk directly, rather than seeking to shift this burden to the other party? How does each party's relative sensitivity to risk influence any compensation given or received as part of a transfer of risk? How does the balance of power in the channel influence the outcome? How significant is the penalty for using risk-neutral thinking in what is fundamentally an exercise in the management of risk?

Unfortunately, most existing models of this setting are ill-equipped for such a task because they either contain no uncertainty at all, or assume risk-neutrality of all decision makers. In the latter case, expected profit (or cost) is the exclusive currency used to measure individual preferences, and transfers of this currency between parties sometimes are described as a "sharing" of risk. But any logic that fails to differentiate between certain and uncertain payoffs fundamentally is at odds with the notion of sensitivity to risk, and therefore may offer spurious recommendations.

The objective of this research is to address this shortcoming by analyzing formally how sensitivity to risk affects both sides of a manufacturer-retailer relationship under various scenarios of relative strategic power, and how these dynamics are altered by the introduction of a manufacturer return policy. To model a channel facing stochastic demand and comprising multiple independent decision-makers, we extend a game-theoretic formulation of Padmanabhan and Png (1997). Our research generalizes theirs (and most of the other existing research on manufacturer return policies) in two key ways. The first is in considering the case in which the retailer is the strategic leader in the channel, in addition to their premise that the manufacturer dictates the distribution policy. Juxtaposing these two scenarios provides the ability to comment on the interaction of channel power and risk attitudes. The second is in the explicit acknowledgment of sensitivity to risk, represented by mean-standard deviation value functions that have been used extensively in a number of literatures.

Examples of meaningful managerial insights that result from this study include the following:

• The penalty for errors in estimating a channel partner's sensitivity to risk can be substantial. The issue is problematic since inferring another firm's risk attitudes could be challenging. While this would be a concern even if the problem was merely one of estimating an unknown, our analysis suggests that the party with less leverage in choosing the channel policy has incentive to feign risk sensitivity by whatever means possible. This can favor the policy that avoids this informational issue.
• Risk sensitivity leads to behaviors that can differ qualitatively from those predicted by risk-neutral analysis. For example, if a channel-leading manufacturer is sufficiently averse to risk, the unit wholesale price charged might actually be lower under a return policy than when no returns are allowed. This is a risk sensitive manufacturer's response to the profit variability induced by retailer returns.
• Being risk sensitive does not always make a party worse off. Net gain might result from how the other party compensates for its partner's reaction to risk.
• A powerful channel member does not automatically prefer to offload channel risk on its trading partners. In particular, a risk sensitive retailer does not always prefer the right to return excess product for full credit, and a risk sensitive manufacturer does not always oppose this. While this is true when the wholesale price is held fixed, it is not true in general.
• In choosing from among potential channel partners a firm should consider their relative risk attitudes, to the extent that these attributes may be ascertained. For instance, a weak retailer should seek manufacturers that are more sensitive to risk if they offer return privileges. Such manufacturers will tend to be more accommodating on the wholesale price.

These findings, which clearly could not arise from any analysis premised on risk neutrality, can serve as the basis of empirical investigation about an important set of issues in the management of distribution channels. They provide more realistic conceptual guidance to channel managers in determining when and how to use return policies, and to students of industrial organization seeking a richer picture of why these institutional practices might appear in some settings but not others.