Intangible capital and productivity: an exploration on a panel of Italian manufacturing firms

The aim of this paper is to investigate the relationship between intangible capital and productivity performance for an unbalanced panel data sample of about 14,000 Italian manufacturing firms over the 18 years 1982-1999. We take advantage of detailed (and rarely available) accounting information on bookvalues and investment flows for different components of intangible capital in an attempt to assess its productivity, both relatively to that of tangible capital and in absolute terms. Contrary to most studies in the existing empirical literature, we do not only rely on an extended Cobb-Douglas production function in which intangible capital and tangible capital enter multiplicatively, which implies that the output elasticities of these two types of capital are constant and that their elasticity of substitution is one. We confront this specification to another Cobb-Douglas formulation in terms of an additive expression “total capital” (i.e., a weighed sum of its tangible and intangible components), which implies a constant marginal productivity of intangible capital relative to tangible capital (but varying output elasticities and an infinite elasticity of substitution). This formulation, which is also often considered, leads to estimates that can be difficult to reconcile with the estimates obtained with the more usual fully multiplicative Cobb-Douglas formulation (in part because depending more crucially on the measurement of intangible capital and tangible capital they are usually less robust – see for example Mairesse and Sassenou, 1991 ). We thus also experiment with the “simplest” encompassing formulation, in which “total capital” is expressed as a CES function of its tangible and intangible components (implying also a constant elasticity of substitution, but possibly different from being one or infinite). Besides confronting different functional forms, we systematically investigate the empirical evidence in the time-series dimension of our data, which allows to control simply for permanent forms of unobserved firm’s heterogeneity at the cost of usually fragile estimates, and not only in the cross-sectional dimension, where they are typically much more robust (and frequently more plausible as well). We do so mainly by considering the traditional standard pooled total, within-firm and one-year and five-year difference estimates, but also various GMM type estimators in an attempt to control for potential errors in variables and simultaneity specification biases. We also further explore the robustness of our main results by comparing them when using alternative measures of intangible and tangible capital as well as different samples or subsamples. Although, as could be expected, all our results cannot be reconciled in full details, they convey overall a rather consistent and reasonable picture. In particular, the time series type evidence is not strongly at variance with the cross-sectional findings, and provides economically and statistically significant estimates of the order of magnitude of the elasticity of intangible capital and the range of its marginal productivity relative to that of tangible capital. Going one step further, we also investigate between potential productivity differences between the different components of intangibles as they are recorded in the firms’ detailed balanced sheets or “capitalised” by us on the basis of the corresponding investment flows (recorded in firms’ current accounts), and between what we can call intellectual capital (i.e. R&D expenditures; patenting and related costs;...) and customer capital (i.e., advertising expenditures; trademarks and related costs;…). Although our results have still to be investigated as thoroughly as when only considering intangible capital as a whole, they seem promising. On the whole, in spite of the great difficulties of measurement and contrary to the scepticism shared by many firm analysts and economists, our findings clearly indicate that firms’ accounting data on intangible capital stocks has real informative content. This should comfort the position of who do consider that intangible expenditures should be and can be realistically treated as investments in practise (not only in theory) and who advocate for the development and progress in measuring and reporting such expenditures and in capitalising them in firms accounts--- and of course in national accounts as well.