Accuracy, quality and reliability of productivity measures

19.141    Economic statistics may be fit for one purpose but may not be fit for others. MFP measures are developed for conducting analysis of long-term productivity growth. It is not ideal for users to employ them for assessing short term productivity fluctuations. Caution needs to be exercised in interpreting productivity measures, which are derived as a 'residual', and are therefore subject to any errors in the output and input measures. Such errors carry a relatively greater importance with respect to productivity estimates, which are calculated as a ratio of outputs to inputs. In the short to medium term, MFP estimates are subject to variations in capacity utilisation or other factors such as the weather. Taking into account these factors, MFP estimates are probably most useful when viewed as average growth rates between growth-cycle peaks, which tends to also remove much of the random error.

19.142    The approach taken for estimating MFP is based on neoclassical economic theory using a translog production function in conjunction with two assumptions: constant returns to scale; and that the marginal products of capital and labour are equal to their respective real market prices. This forms the basis of the growth accounting approach to estimating MFP.

19.143    However, these assumptions are unlikely to hold in practice. If there are scale efficiencies, then this will also be captured as an increase in MFP. This possibility is likely as there would be many firms operating in an environment of increasing returns to scale, especially over short periods. The assumption that the marginal products of capital and labour are equal to their market price is based on the existence of perfect competition in factor markets.

19.144    In practice, growth in MFP may contain the impact of many phenomena in addition to disembodied technological change, such as:

• economies of scale and scope;
• reallocation effects of capital and labour; 
• changes in the work force and management practices;
• climate and water availability;
• variations in capacity utilisation; and
• measurement error.

19.145    Also, MFP estimates are subject to the vagaries of the growth in the business cycle (as capacity utilisation varies so does MFP growth). Taking into account all of these factors, MFP estimates are probably most useful when computed as average growth rates between growth-cycle peaks, which are determined as peak deviations of the market sector MFP index from its long-term trend. In this way, most of the effects of variations in capacity utilisation and much of the random error are removed. However, average growth rates still reflect any systematic bias resulting from the methodology and data used.