Measurement of productivity growth in U.S. manufacturing

1 Monthly Labor ReviewJuly 1995 13 Measuring PrMeasuring PrMeasuring PrMeasuring PrMeasuring productivity Groductivity Groductivity Groductivity Groductivity GrowthowthowthowthowthThe indexes of multifactor productivity fortwo-digit manufacturing sectors pre-pared by the Bureau of Labor Statisticsfor sectors in manufacturing have been revisedand extended to cover the 1949 92 period. Theseindexes, also called the KLEMS multifactormeasures, compare changes in output to changesin a composite of all the inputs used in produc-tion capital, labor, energy inputs, nonenergymaterial inputs, and business Becauseof this comprehensive input list, these indexesgive an indication of advances in technology andproduction efficiency in these broad sectors, im-portant topics as the economy emerges from therecession of the early 1990 article discusses the Measurement ofmultifactor productivity for manufacturing andanalyzes growth trends within the the years, a wide variety of productiv-ity statistics have appeared in the literature.

2 Dis-tinguished by the concepts underlying the mea-surement of output, the methods of aggregation,and the inputs included for analysis. Recent ad-ditions of superlative indexes of gross domes-tic product (GDP) by industry to the NationalIncome and Product Accounts, prepared by theBureau of economic Analysis, Departmentof Commerce, have enhanced available alterna-tives for measuring manufacturing changes in the way BLS measures manu-facturing productivity are also productivity growth trends arethen examined for the overall manufacturingsector and for 19 two-digit SIC manufacturingMeasurMeasurMeasurMeasurMea surement of prement of prement of prement of prement of productivityoductivityoductivityoductivi tyoductivitygrgrgrgrgrowth in manufacturingowth in manufacturingowth in manufacturingowth in manufacturingowth in manufacturingProductivity Measurement cannot be restrictedto capital and labor factors intermediate inputsconstitute too large a part of the cost structure.

3 Revised and extended data show upward trendin multifactor productivity growthW illiam GullicksonW illiam Gullickson is aneconomist in the Officeof productivity andTechnology, Bureau ofLabor In particular, early postwar andmore recent productivity growth trends are com-pared. When this comparison was last discussedin 1992, data were available through 1988, cov-ering a period of rapid growth following emer-gence from the 1982 Because of thisgrowth, multifactor productivity growth seemedto have regained much of its early postwar mo-mentum. It is now possible to examine recenttrends more comprehensively because the ex-tended series cover the 1990 business cycle peak,the brief recession that followed in 1991, and arecovery period in 1992.

4 These trends indicatethat the productivity growth rates of the earlypostwar period were not entirely regained dur-ing the 1980 in measurIssues in measurIssues in measurIssues in measurIssues in measurementementementementementUntil recently, the Bureau of Labor Statisticsproduced two distinct and fairly different mea-sures of multifactor productivity for the manu-facturing sector. One measure was a comparisonof net output to capital and labor Theother was the KLEMS multifactor measure, issuedalong with multifactor productivity measures forbroad (two-digit SIC) manufacturing industries,which compares sectoral output to capital, la-bor, and intermediate the future, BLS will use the measure basedon sectoral output for both purposes, while con-tinuing to use a somewhat modified net output-type multifactor productivity measure for its in-ternational comparisons of multifactor produc-tivity.

5 Some further background on measures14 Monthly Labor ReviewJuly 1995 Measure productivity growthThe Tornqvist gross output is obtained bychain-weighting four-digit industry deflated values of ship-ments to the two-digit level, at which level inventorychange is added; then aggregating to total of shipments are from the Bureau of the Cen-sus and are maintained, among other four-digit industrydata, in BLS; deflated shipments for each industry are equalto the census value of shipments adjusted by a four-digitdeflator which in turn is a composite of five-digit deflatorsfrom the BLS producer price program, the Bureau of Eco-nomic Analysis, and other sources. All data are arrangedaccording to the 1987 Standard Industrial Classification(SIC) system.

6 A Tornqvist aggregate quantity index is achain of two-period indexes (Q2/Q1), each of which is ob-tained as a weighted average of individual item indexes,with weights based on cost or value shares taken from thetwo periods. For further discussion of Tornqvist indexes,see Trends in Multifactor productivity , 1948 81, Bulletin2178 (Bureau of Labor Statistics, 1983).The Tornqvist sectoral output measure. Sectoral outputis the name given to gross output less intrasector transac-tions. This name was introduced by Frank M. Gollop. See Accounting for Intermediate Input: The Link BetweenSectoral and Aggregate Measures of productivity growth , in Measurement and Interpretation of productivity (Wash-ington, National Academy of Sciences, 1979), pp.

7 318 output thus represents deliveries to consumers out-side the industry. The BLS sector output index is a Tornqvistindex, obtained by removing estimated intrasector transac-tions from the Tornqvist gross output measure using cur-rent weights for the output for an industry represents deliveries toAlterAlterAlterAlterAlternative output measurnative output measurnative output measurnative output measurnative output measuresesesesesconsumers outside the industry; for total manufactur-ing, sector output represents deliveries to consumersoutside manufacturing . Intrasector transactions are es-timated from input-output tables published by the Bu-reau of economic Analysis in benchmark years, and pre-pared by BLS for other gross measure ofgross output underlies the estimation of benchmark-years-weighted gross product originating; the latter series is pub-lished in the National Income and Product is designed to achieve mov-ing weights even when all the requisite data are not avail-able annually.

8 See Allan H. Young, Alternative Measuresof Change in Real Output and Prices, Survey of CurrentBusiness, April 1992; and Jack E. Triplett, EconomicTheory and BEA s Alternative Quantity and Price Indexes, Survey of Current Business, April the National Income and Product Ac-counts, but especially in manufacturing , more detaileddata are available in the years of the economic censuses,usually 5 years apart. The benchmark-years-weightedquantity index is the (geometric) mean of two fixed-weighted indexes a Laspeyres index, based on pricesof the first benchmark year and a Paasche index, basedon prices of the second. The averaging of Paasche andLaspeyres indexes yields a Fisher Ideal index, whichhas among its benefits the quality of reversibility : anytwo of the benchmark-to-benchmark ratios of quantity,price, and value suggest the third.

9 The benchmark-years-weighted gross output series are available for two-digit manufacturing industries from 1977 to 1987, ac-cording to the 1972 help explain these changes are provided in the follow-ing sections. The main issues have to do with which inputsand outputs should be included in a multifactor productivityratio and how heterogeneous inputs and outputs should beweighted is engaged in efforts to insure that itsmeasures conform, as nearly as possible, to some basic prin-ciples of productivity Measurement which have been devel-oped in the economics One of the basic principlesis that inputs be as comprehensive as possible, so that pro-ductivity growth does not merely reflect changes through timein unmeasured inputs.

10 Thus, the multifactor productivitymeasures for manufacturing industries presented later in thisarticle take into account all intermediate inputs (energy andother materials and business services). The importance of in-termediates first gained prominence in the literature becauseof the events of the 1970 In the period following the OPECoil embargo, fuel prices rose almost 150 percent in a 4-yearperiod (1973 77) and researchers began to suspect that thisprice increase was contributing to the emerging slowdown inproductivity growth . Since then, increases in the use of busi-ness services, such as equipment leasing, computer services,and the use of temporary labor all of which could have animportant impact on production and employment have af-fected productivity second principle is that inputs and outputs be defined ascomprehensively as possible without double counting.