Transition and path-dependence in knowledge-intensive industry location: Case of Russian professional services

This paper searches for roots of current spatial pattern of professional services in location of Soviet-era R&D sector. The Soviet economy sponsored massive R&D oriented mainly on military purposes. Research was carried out in large institutions affiliated with academia or industry. After the collapse of socialism, military spending and related R&D decreased dramatically. Many researchers left Soviet-style institutions and succeed in the market economy. Did however this process mean anything for geography? Under the centrally planned economy, locational decisions were driven by non-market motives. Moreover, it was near-impossible for planners to anticipate which regions would be more promising under market. I focus on professional services since these industries do not rely on physical capital endowments, so sunk costs are unlikely to impose path-dependence. Professional services in Russia typically hire young educated persons which are expected to be mobile, so imperfect labor market is also unlikely to stick people to places ? unlike human capital externalities which are plausibly to do so. I regress employment in professional services in 2009-2011 by 76 Russian regions on the number of R&D staff in 1991. I consider three industries: architecture and engineering; information technology; accounting, auditing and management consulting. Controls to capture industry location fundamentals include modern-day gross regional product or overall employment, number of employees with university degree, number of R&D staff and urbanization. It should be noted that present-day number of R&D staff comprises primarily employees of state-owned Soviet-style institutions while professional service providers are typically up-to-date privately-owned firms. Results reveal that employment in Soviet R&D positively and significantly affects present-day employment in engineering and in IT. No such evidence is found for accounting, auditing and management consulting in which industries researchers' skills were less relevant. Results remains when shares in employment of professional services and R&D sector are plugged into regression instead of absolute numbers. I consider different explanations for this phenomenon. I attempt to track influence of 1991 R&D-related employment on current output-per-worker in IT and engineering and find positive correlation, although marginally significant. So, it is unlikely that Soviet-era pools of human capital created regional poverty traps due to low migration rates, and human capital externalities theory seems plausible. I also find that regions with greater number of R&D staff in 1991 now have greater number of SMEs both in business services and in the rest of economy, so entrepreneurship is a likely mechanism to impose path-dependence.