Do We Really Need Numerous Observations to Select Candidates? (The d-Day Theorem)

Abstract

Under a realistic assumption of imperfectness of the information, most organizations have to run their internal promotion process under the double constraint of minimizing information costs and maximizing the principal’s confidence about the candidate who will be finally promoted. However, if it is straightforward that concerning the first aim, it is more or less always possible to observe the candidates in their current duties without any supplementary cost (except that of control), it is less obvious that these observations can constitute a relevant information insofar as the promotion-maker must know how to use it in a consistent way. Accordingly, we want here to insist on two particular characteristics of real promotion decisions: (i) the undeterministic nature of the workers’ performances and (ii) the subsequent principal’s information process. It is obvious that, from a practical viewpoint, any individual’s ability to produce, whatever the organization activity, is depending on outside factors which stochastically twist this genuine ability, such as mood, illness or worries... Hence, the principal can be considered as unable to deterministically focus on the best candidate for promotion (i.e. with a single observation). Therefore, he must design an information process which helps him to decrease the uncertainty attached to the observations. In the literature, such a process is generically called a contest, i.e. a set of independent observations corresponding to the agents’ performances which allows the principal to compare between them (for further details on the seminal concept, see e.g. Lazear and Rosen 1981, Meyer 1991). Intuitively such a framework recalls some of well-known situations in social choice theory: since the workers are here viewed as strategically passive, they can be theoretically treated as ‘alternatives’ and their daily performances can be assimilated to a set of numerous rankings which can be analyzed as if they were numerous individual preferences (see, e.g. Moulin 1988). Moreover, a promotion is here a closed concept to that of a standard choice decision: the principal must decide according to a global information based on the daily observations exactly as the society must choose according to the set of the individuals’ preferences. The link between these two a priori disjoint domains is here deeply exploited, especially for the second result which can be understood as an ‘Arrovian Theorem’ applied to organizational economics.