As a prelude to an examination of productivity trends over time, this section considers various methods of measuring the output and input components of productivity ratios and some of the difficulties and limitations of the resulting estimates.

With respect to output, ideally the numbers of units of each category of tangible commodity or service should be counted in successive time periods and aggregated for the firm, industry, or total economy in terms of some indicator of relative importance, usually price or cost per unit as of a particular period. The unit value “weights”—price, cost, or other—must be held constant for two or more periods being compared so that changes in aggregate output reflect changes in physical volumes rather than in prices. An alternative procedure that produces the same results with ideal data is to “deflate” current values of the various items produced by index numbers that reflect relative price changes in order to eliminate the effects of price changes. Price deflation is usually employed to obtain estimates of real gross product by sector and industry to be used as numerators of productivity ratios. For tangible industrial production measures, quantities of the various commodities are generally weighted together by constant unit values.

Unfortunately, in most countries data on quantities and prices for many outputs of the finance and service industries are deficient. In the broader real gross product estimates, changes in outputs of a portion of such services are approximated by estimating changes in inputs. Estimates so derived are not suitable for productivity measurement, however. They impart a downward bias to estimates of real product and productivity for the services sector and its affected components and hence for the economy as a whole.

Other problems in estimating output arise in adjusting estimates of outputs to take account of quality change, measuring quantities or prices of nonstandard custom-made products, and estimating outputs of nonmarket goods and services. Partial adjustments for quality changes may be made when increases in real costs per unit are associated with the improvements. But it is generally agreed that physical-volume or real-product measures fail to capture at least part of the improvements in product quality, as distinguished from relative shifts among alternative qualities (price-lines) of a given product. Methods of estimating changes in the physical volume of custom-built products, such as buildings or other major structures, have improved in recent years. But changes in the output of nonmarket goods and services, such as those of governments, households, and nonprofit institutions, must generally be measured by changes in inputs. In consequence, productivity estimates are usually confined to the predominant business (enterprise) sector of an economy.

Labour input is relatively easy to measure if one is content to count heads of persons engaged in production or, preferably, hours worked. But in fact, the available hours data often relate to hours paid for, rather than hours worked, and these tend to rise in relation to hours at the workplace as the number of paid holidays and leaves are increased. Official estimates generally do not differentiate among various categories of labour. But some academic economists measure labour inputs by occupation and/or industry and possibly other categories and weight the aggregate in each category by a measure of the average compensation in some designated base period. As the average levels of education, training, skills, and experience of workers increase, the weighted measures rise relative to unweighted measures of labour input. Change in the ratio of the two indicates change in the quality of labour input, which is an important part of the explanation of change in productivity.

Capital input is usually assumed to change in the same direction as and proportionally to changes in the real stocks of structures, equipment, inventories, and natural resources. The rates of return on those capital goods in some base period are taken to be indicative of their productivity for the purpose of weighting them together with other factor inputs. Some analysts adjust the capital estimates to take into account changing rates of utilization of capacity; otherwise, changes in utilization rates are reflected in the productivity estimates.

Interindustry purchases and sales of intermediate products—those materials, energy, and other services that are consumed in the production process—are accounted for on a value-added basis and cancel out in the national income and product estimates by industry (one industry’s output being the next one’s input). But if intermediate purchases are counted as an input for comparisons with gross output estimates, they are measured in the same manner as described for outputs.