NON-FRONTIER MEASURES OF EFFICIENCY, PROGRESS AND REGRESS FOR TIME-SERIES DATA

In the classical literature on efficiency measurement, the notion of e fficiency is intimately linked with the one of production frontier: sp ecifically, a productive activity is called efficient if the vector of input-output quantities that describes it belongs to the (efficient) frontier of some appropriately defined production set; and the activit y is called inefficient if this vector belongs to the interior of the set. Moreover, when time-related data are involved, a sharp distinctio n is usually made between shifts of the frontier and gains in efficien cy. Running against this tradition, we deal in this paper with efficie ncy, progress and regress without using the frontier concept. For desc ribing the relation between inputs and outputs in production, we subst itute another notion called ''benchmark production correspondence'', w ith respect to which we suggest that progress and regress be measured without having to distinguish between progress and efficiency gains, o r between regress and efficiency losses. An illustration with a monthl y time series covering almost 15 years of the activities of an urban t ransit company is presented and contrasted with DEA and FDH non-parame tric frontier approaches.