Presentation on theme: "Output and productivity measurement in a US/UK comparison,"— Presentation transcript:
1Output and productivity measurement in a US/UK comparison, 1900-1950 Output and productivity measurement in a US/UK comparison,Reconciling sectoral benchmarks and time seriesPieter Woltjer & Herman de JongUniversity of GroningenFaculty of Economics and Business
2IntroductionLong-span projections criticized through confrontations with benchmarks (Ward & Devereux, 2003)Sectoral, industry-of-origin, studies suffer from the same problems (Broadberry & Burhop, 2007)For pre-war period some benchmarks are available to cross-check time series projectionsMostly based on quantity approach (Rostas, 1948)Data is available to apply ICOP unit value approach (Maddison & van Ark, 1988)Recent studies stress advantages of ICOP approach over quantity based comparisons (Fremdling et al., 2007)
3OutlineExamine available benchmark/time series evidence for first half 20th century for the US and the UKDiscuss differences benchmarking methods, quantity vs. unit value approachAnalyze comparative productivity using time series and shift-shareCompare benchmark and time series, is reconciliation feasible?
4Benchmark comparisons US productivity , 1935 (UK =100)IndustryRostasde Jong & Woltjer(quantity)(unit value)Food, Drink and Tobacco207152Textiles and Clothing146174Chemicals213263Metals200185Engineering283327Miscellaneous260247Total Manufacturing224Sources: Broadberry & Crafts (1992); Rostas (1948); de Jong & Woltjer (2010)
8Reconciliation time series and benchmarks Reconciliation time series and benchmarksAre the results from benchmarks and time series consistent, or transitive?Transitivity across time and space not realistic or reasonable to impose on international comparison (Dalgaard & Sorensen, 2002)Intransitivity error will increase if individual price movements are large (length of period) and if sectoral structure of countries under comparison differs (Szilagyi, 1984)Sharp changes in price level compared to the base country tend to worsen intransitivity error (Aten & Heston, 2002)Early 20th century could expect large deviations (wars, inflation)
11Reconciling time series and benchmarks Reconciling time series and benchmarksDiscrepancies of projections will depend largely on the accuracy of underlying benchmarks and time seriesBenchmarks are affected by measurement errors (e.g. quality of products) and representativeness issues (e.g. selection bias)ICOP approach represents substantial improvement over older quantity based international comparisonsTime series suffer from measurement errors (e.g. technological development) and international inconsistencies (differences in national accounting)New historical national accounts rely on more systematic methodology, aids international comparisons
12Conclusions For the moment discrepancy between both methods remains ConclusionsFor the moment discrepancy between both methods remainsWe argue that there is a clear need for more (ICOP) benchmark studies for early 20th centuryIn addition, future research should put more emphasis on the link between benchmarks and long-span projectionsIntransitivity and measurement errors can be quantified by analyzing shifts in international and inter-temporal weightsHistorical national accounts should be revisited, and supplemented by new, basic, unchained time series based on price data from direct benchmark comparisonsHistorical national accounts stress the comparability of methodologies and coverage over time as well as between countries