Adaptive pricing policy for a job-order production system

Abstract

This thesis deals with the establishment of a routine pricing policy for a job-order production system where job specification does not differ much, order by order, and thus it is appropriate to employ a routine pricing policy rather than a one-shot decision. Two types of models are formulated, a single-period and a multi-period model. It is suggested that the final pricing decision may be made in combination of both the solutions from the single-period and the multi-period model, taking into account of the risk in relying upon one or the other model only. The model is extended to the adaptive decision rule by employing adaptive smoothing technique. Adaptive smoothing technique is typically applied for forecasting some state variables in a very-short-term model, e.g., for forecasting demand in inventory problems. In this study, however, the technique is applied for ``forecasting`` the model parameters themselves. Initially the values of the parameters would be estimated by regression technique. However, after the initial setting of the model, the parameter values may be established for each new period by appropriately forecasting them with adaptive smoothing technique. The smoothing technique which will be adopted in this study is actually based upon the regression technique itself. In usual forecasting models, the smoothing is done between ``actual values`` and the smoothed values of some state variables. Here, the smoothing will be done between the ``estimated values obtained by a new regression`` and the ``values smoothed along the old regressions``. Finally, the models developed in this study are applied to real situations.