The purpose of the paper is to analyze the effects of advertising and R&D expenditures on the values of intangibles for the 1,091 Japanese manufacturing firms during the 38 years from 1965 to 2002. The accounting data exhibits an unbalanced panel structure. The paper employs a general linear regression model with parameters varying over time, which leads a generalized least squares (GLS) method of estimation instead of conventional ordinary least squares (OLS). Empirical results reveal that (1) the effects of the advertising expenditure does not change during the sample periods except for the bubble years, (2) the effects of the R&D expenditure increases in the early 80s, decreases from 85 to 98, and increases again since 2000, (3) the effects of the R&D expenditure are larger than those of the advertising expenditure, and (4) there exist strong autocorrelations for the error terms, which justify the use of the GLS instead of the OLS.
View full abstract