This paper tests New Classical and Keynesian explanations of output determination within an encompassing "factor utilization" model wherein the output decision by producers is modelled as the choice of a utilization rate for employed factors. In this encompassing model, the ratio of actual to normal output (with the latter defined by a nested CES vintage production function with capital, energy and employment as factor inputs) is explained by unexpected sales (a Keynesian element), abnormal profitability (one component of which is the Lucas "price surprise" effect), and abnormal inventories. Results using Canadian data show that the Keynesian and New Classical elements contribute explanatory power, as does the production-function-based measure of normal output, while each of these partial models is strongly rejected in favour of the encompassing model. The highly structured factor utilization model is also seen to fit better than an unstructured VAR model. U.S. data confirm the results, and show that there are significant effects from abnormal demand, profitability and inventory levels even if the labour and capital components of normal output are defined using hours and utilized capital rather than employment and the capital stock. The results are also confirmed using alternative output (and hence input) concepts, using a translog function instead of a CES function to define normal output, and using data for several other major industrial countries.
