A Generalized Exactly Additive Decomposition of Aggregate Labor Productivity Growth

Aggregate labor productivity (ALP) growth--i.e., growth of output per unit of labor--may be decomposed into additive contributions due to within-sector productivity growth effect, dynamic structural reallocation effect (Baumol effect), and static structural reallocation effect (Denison effect) of cross-sectional components (e.g., industry or region) of output and labor. This paper implements ALP growth decomposition that is "generalized" to output in constant prices and to output in chained prices (i.e., chained volume measure or CVM) and "exactly additive" since with either output the sum of contributions exactly equals "actual" ALP growth. It compares this "generalized exactly additive" (GEAD) decomposition to the "traditional" (TRAD) ALP growth decomposition devised for output in constant prices. The results show GEAD and TRAD are exactly additive when output is in constant prices but GEAD is exactly additive while TRAD is not when output is in CVM. Also, GEAD components are empirically purer than or analytically superior to those from TRAD. Moreover, considering that contributions to ALP growth can be classified by industry or region each year over many years, GEAD provides a more well-grounded picture over time of industrial or regional transformation than TRAD. Therefore, GEAD should replace TRAD in practice.