The Cost of Digitalization: Digital Tax Administration and Corporate Innovation Crowding-out

While digital technologies in tax administration help local governments meet revenue collection goals, their impact on corporate innovation remains contested. This study unpacks this relationship by testing two competing theoretical logics against each other—a resource-based crowding-out logic and an agency-based incentive logic. Leveraging the Golden Tax Project III (GTP_III) as a quasi-natural experiment and using a time-varying difference-in-differences design on Chinese A-share listed firms from 2010 to 2021, we empirically determine which logic dominates. The results reveal a striking asymmetry. GTP_III robustly crowds out Research and Development (R&D) expenditure by increasing corporate tax burdens and financial risks. The incentive channel, which was expected to operate through reduced agency costs and inhibited rent-seeking, fails to materialize to a meaningful degree. This crowding-out dominance extends across the entire innovation lifecycle: the decline in R&D inputs is not offset by efficiency gains. Total factor productivity declines and innovation becomes more intermittent. We argue that this asymmetry reflects specific institutional conditions—intense local government fiscal pressures combined with uneven market-supporting institutions—that channel enhanced monitoring capacity toward extraction rather than governance enablement. These findings caution against optimistic assumptions that digital governance automatically promotes corporate innovation and call for policy designs that strategically balance revenue collection with the institutional conditions needed to activate governance dividends.