³ÉÈË´óƬ

Three styles of explanation have been advanced by economists seeking to account for the so-called "productivity paradox". The coincidence of a persisting slowdown in the growth of measured total factor productivity (TFP) in the US, since the mid-1970's, with the wave of information technology (It) innovations, is said by some to be an illusion due to the mismeasurement of real output growth; by others to expose the mistaken expectations about the benefits of computerization; and by still others to reflect the amount of time, and the volume of intangible investments in "learning", and the time required for ancillary innovations that allow the new digital technologies to be applied in ways that are reflected in measured productivity growth. This paper shows that rather than viewing these as competing hypotheses, the dynamics of the transition to a new technological and economic regime based upon a general purpose technology (GPT) should be understood to be likely to give rise to all three "effects." It more fully articulates and supports this thesis, which was first advanced in the "computer and dynamo" papers by David (1990, 1991). The relevance of that historical experience is re-asserted and supported by further evidence rebutting skeptics who have argued that the diffusion of electrification and computerization have little in common.