This paper provides a decomposition of productivity growth into its components, namely technical change, returns to scale, and disequilibrium effects of quasi-fixed factors, within a complete theoretical framework for measurement and attribution using the parametric approach. It also provides an empirical application of the framework to Greek agriculture, which has exhibited a productivity slowdown since the late 1970s. The empirical results indicate that technical change and disequilibrium effects are the main factors affecting productivity change in Greek agriculture. The contribution of the scale effect, although positive, has been quite small. The slowdown in productivity growth in Greek agriculture since the late 1970s is attributed mainly to the market disequilibrium effect of capital induced by the decline in gross capital investment.Although both aspects of productivity analysis, i.e., measurement accuracy and attribution, are desirable, the choice of the particular approach to be followed in an empirical analysis depends, also, on data availability, computational simplicity, and a priori restrictions imposed in the empirical model. The necessary data are more or less similar for all approaches. In terms of computational simplicity, the accounting approach is the easiest to implement, regardless of the number of inputs and outputs included in the analysis, whilst the complication of the non-parametric approach is positively related to the dimension of production structure in the output and input space (Diewert, 1980). In contrast, the ranking of the three approaches is reversed in terms of a priori restrictions imposed in empirical modelling. That is, most severe are the underlying assumptions of the accounting approach, whereas the non-parametric approach is more flexible in terms of functional specifications. Hence, in terms of the last two criteria, the parametric approach lies somewhere between the other two, but it has strong advantages in terms of analytical capability.Recent literature on TFP shows that considerably more research effort has been devoted to the parametric approach, implying that the computational effort required for the econometric estimation of such models has been considerably reduced with the use of flexible functional forms and duality. Nevertheless, there are certain concerns about the sensitivity of the parametric approach to alternative structural and behavioural assumptions (Callan, 1991).Accurate measures of TFP and reliable analysis of its trend can be obtained within the parametric approach only if the appropriate structural and behavioural assumptions are correctly specified. Structural assumptions of production refer to returns to scale, technical change, productive efficiency, and capacity utilisation. On the other hand, behavioural assumptions refer to the objective function of the firm, market disequilibrium effects, market structure, and the importance of various (economic and environmental) regulations.* Denny et af., (1981) established the role of technical chang...