When economic growth is decomposed by growth accounting into
contributing factors, technological progress is measured by total factor
productivity. But, the measured primal TFP coincides with the technological
progress only under the strict assumptions of perfect competition and constant
returns to scale (CRS). In this paper, we first measure dual TFP which does
not depend upon the CRS assumption. We also measure Hall type TFP which
relaxes the assumption of perfect competition. The former is measured to have
higher variability than the primal TFP, while the latter has lower value. When
we relax both assumptions, the estimates of the technological progress turn
out to be negligible while returns to scale appear to have increased since
1990s. While the increasing returns to scale (IRS) in the total economy might
be interpreted as evidences of positive externality and spillover effects due to
the transition to knowledge-based economy, the strong IRS in service industry
may suggest the role of increased size of firms as well as franchising.