In most developing countries data limitations lead to the use of assumptions that compromise studies on the measurement of capital and its impact on productivity analysis. A possible approach is to use the ratio of the value of tractor sales to overall expenditure to impute overall machinery sales. The use of a constant ratio over an extended period results in increasingly incorrect estimates and fails to reveal the changing nature of mechanisation. In this paper, the probems with such an approach are highlighted through an analysis of the historic share of tractor sales to overall machinery sales in South Africa. This paper establishes that the current methods lead to underestimation in the overall value of machinery and implements sales in South Africa by approximately a $100 million per annum in recent years. An alternative method is suggested and the implications of a new capital formation series are discussed.