This report comprises a number of related components focussed on the role of capital and capital formation in production, productivity and competitiveness in Canadian agriculture, concentrating on the Prairie region of Western Canada in the period 1970 to the early 1990s. The report includes the analysis of investment flows and capital stocks data, problems in measuring productivity, and an assessment of the relative roles of technological change and economies of scale in estimated productivity growth. In addition to the nature and extent of technical change, other features of the structure of production technology are analyzed by estimation of a translog cost function. Input mix and investment data indicate an important increment in investment, resulting in a larger machinery stock in the 1970s and a modest increment in the stocks of capital related to agricultural land and buildings. Changes in the input mix result from the initial post-war substitution of capital for labour and an eventual rising share of materials inputs including agrichemicals. The land input appears to change little and slowly. Land related investment and repairs also show a slow but steady growth over the period. In the 1980s, the labour share slowly rises, reversing its previous downward trend. The important increase in capital-related investments, particularly machinery, in the 1970s is followed by a considerable drop in the 1980s. Dis-investment appears to be taking place in the late 1980s and early 1990s. However, productivity and total output do not appear to be shrinking, and productive capacity does not appear to have been seriously affected in the short run by the lower levels of investment in the 1980s. Cost function parameters, elasticities of derived demand for inputs, and Allen partial elasticities of input substitution are estimated in the report. Results indicate an increasingly rigid production structure in the 1970s and early 1980s, lower elasticities of derived input demand and reduced input substitution. Such trends suggest that technology in prairie farming was becoming a "technological package". This trend was modestly reversed in the late 1980s when a slowly changing input mix resulted from reduced use of chemicals and machinery. These changes may be related to changing economic conditions which generally favoured cost reduction as opposed to output expansion. The relatively slow response indicates the persistence of rigidities. It is not yet clear whether slightly lessened rigidity is the start of a new trend or only temporary. Testing functional properties of the cost function indicates rejection of homotheticity, of constant return to scale, and of Hicks neutrality. The index number methodology for the empirical measurement of agricultural productivity is analyzed. A major problem is measurement of "durable" capital items. Aggregation or indexing procedures is another important conceptual issue. Given the conceptual superiority of flexible indexes we would recommend that Divisia-based or chained Fisher indexing be employed rather than traditional indexes such as the Laspeyres or Paasche. Our calculations suggest there is little practical difference in productivity estimates based on the Tornqvist-Theil approximation to the Divisia index as opposed to the Fisher "ideal" chained index. Total factor productivity, terms of trade and return to costs ratios are estimated. Although productivity growth nearly fully compensated for adverse movements in the terms of trade for prairie agriculture over the entire postwar period from 1948 to 1991, it was much less effective in doing so in the 1980s and profitability deteriorated. The evidence suggests that the farming system in Western Canada has been experiencing important transformations in the last two decades in our period of study. In terms of the structure of production technology, our fmdings indicate non-homotheticity, biased technical change, and a more important role for economies of scale. For productivity measurements, the use of flexible forms, such as Divisia or Fisher Chained procedures, is preferred. The agricultural system, having achieved a new ceiling in investment in the 1970s, went through a process of adjustments and correction in the second half of the 1980s. There was a reduction in the annual level of investments and a shrinkage in the capital stock, mostly due to a decline in farm machinery. Given technical change and data problems, it is difficult to say if the level of investment prevailing at the end of the period was sufficient to compensate for actual capital depreciation. Future data availability and further research may provide a more definite answer.