The rebound effect: Microeconomic definitions, limitations and extensions

The rebound effect results in part from an increased consumption of energy services following an improvement in the technical efficiency of delivering those services. This increased consumption offsets the energy savings that may otherwise be achieved. If the rebound effect is sufficiently large it may undermine the rationale for policy measures to encourage energy efficiency. The nature and magnitude of the rebound effect is the focus of long-running dispute with energy economics. This paper brings together previous theoretical work to provide a rigorous definition of the rebound effect, to clarify key conceptual issues and to highlight the potential consequences of various assumptions for empirical estimates of the effect. The focus is on the direct rebound effect for a single energy service — indirect and economy-wide rebound effects are not discussed. Beginning with Khazzoom's original definition of the rebound effect, we expose the limitations of three simplifying assumptions on which this definition is based. First, we argue that capital costs form an important part of the total cost of providing energy services and that empirical studies that estimate rebound effects from variations in energy prices are prone to bias. Second, we argue that energy efficiency should be treated as an endogenous variable and that empirical estimates of the rebound effect may need to apply a simultaneous equation model to capture the joint determination of key variables. Third, we explore the implications of the opportunity costs of time in the production of energy services and highlight the consequences for energy use of improved 'time efficiency', the influence of time costs on the rebound effect and the existence of a parallel rebound effect with respect to time. Each of these considerations serves to highlight the difficulties in obtaining reliable estimates of the rebound effect and the different factors that need to be controlled for. We discuss the implications of these findings for econometric studies and argue that several existing studies may overestimate the magnitude of the effect.