Technology overload? macroeconomic implications of accelerated obsolescence

Since the mid-1990s computing revolution, advanced economies have shown a cluster of regularities — a decade-long productivity boom followed by a slowdown below trend, a sharply declining labour share, and falling capital efficiency — and this paper argues they are not isolated but reflect a common structural change, with one possible driver being faster obsolescence of capital in use. Decomposing BEA economic depreciation for 42 non-residential private equipment and software assets into physical and obsolescence components (using a Federal Reserve Board perpetual-inventory approach with hyperbolic decay that explicitly excludes obsolescence), the author constructs an implied US obsolescence rate for 1970–2023 that is roughly flat before 1995 and then trends up: from an average of 4.2% (constant-dollar) and 5.6% (current-dollar) over 1970–1995 to 6.6% and 6.4% respectively by the late 2010s, while the physical rate rises much less. Feeding a one-time rise in the obsolescence rate from 5% to 7.5% — dated just after 1995 — into an endogenous growth model with directed technical change à la Acemoglu (2003), calibrated to US moments and with efficient labour and efficient capital as gross complements (elasticity set to 0.5), produces a temporary productivity boom that fades within about a decade, followed by a new balanced growth path with productivity growth falling from 1.70% to 1.51% (a 0.19 percentage-point reduction) and the labour income share falling from 62.6% to 56.4% (a 6.2 percentage-point decline); the observed US counterparts are 1.47% and 56.9%. The mechanism is that faster replacement makes new capital designs scarcer in efficiency units, raising the relative return to creating capital technologies and drawing resources away from labour skill creation, so that under labour–capital complementarity the labour share settles lower and long-run growth slows; the model also reproduces a wedge between wage and productivity levels of roughly the ~20 log points visible in US data by 2019, and a decline in capital efficiency (measured as the inverse incremental capital–output ratio, which falls about 0.68 log points in the data) though with a slightly smaller magnitude. Sensitivity analysis leaves the qualitative results intact but shows the size of the productivity slowdown is sensitive to the initial allocation of work effort and the size of the labour-share decline to the assumed initial obsolescence rate; a companion panel of 16 advanced economies shows the same productivity, labour-share and capital-efficiency patterns, and the author frames the obsolescence channel as one possible contributing factor that neither contradicts nor dismisses prevailing explanations.

Seda Basihos | Macro Paper Warehouse

Technology overload? macroeconomic implications of accelerated obsolescence