Heterogenous treatment effects of incorporating public innovation intermediaries: Evidence from extension, research, and inventive activities

Abstract

The incorporation of public organizations is implemented by the central and local governments to improve their efficiency, which should enhance their contributions to national and local economies. In Japan, a series of administrative reforms were implemented at the national and local governments since the late 1990s, some of which took the form of the incorporation. Taking an example of the incorporation of technology extension service (TES) providers established by local governments, this study evaluates the average treatment effects on the treated (ATT) by applying the difference-in-differences (DID) model to panel data of TES providers for small and medium-sized enterprises (SMEs), Kohsetsushi. Unlike the uniform and simultaneous incorporation of national universities, the incorporation of Kohsetsushi is at the discretion of local governments and the timings of incorporation vary. Applying the conventional two-way fixed effects (TWFE) model to panel data with staggered treatments may yield biased ATTs. Following Callaway and Sant’Anna (2021), this study adopted their DID model (CS-DID) to avoid forbidden comparisons between late and early treated units. The ATTs in terms of extension activities are heterogenous and significantly positive for TWFE-DID but statistically insignificant for CS-DID. The sources of heterogenous treatment effects seem to be decline in innovation agglomerations that made localized spillover insignificant and unobservable organizational capabilities of local governments to exploit the incentive system reform for geographically broader spillover.