Weberian bureaucracy does not systematically enhance growth! At least, the evidence is not sufficiently clear to conclude on a relationship.

This is the main finding of my paper with Agnes Cornell & Jan Teorell, now published in @cps_journal .

https://journals.sagepub.com/doi/full/10.1177/0010414020912262

Thread:

(2) We use @vdeminstitute data back to 1789 on different features of Weberian bureaucracy, including impartial and rule-following administration, meritocratic recruitment and promotion of bureaucrats, etc. None of these features are clearly related to growth in our analysis.

(3) When using within-country variation over time, the positive correlation from previous cross-sectional studies vanishes. Countries with “better” bureaucracies are richer, but not because bureaucracy causes growth. (Richer countries could be better at building bureaucracies.)

(4) One potential reason for our findings: Weberian bureaucracy enables effective implementation, but governments can use this tool to effectively pursue disastrous policies (or good ones). Capacity is not enough for growth, you also need incentives to pursue the right policies.

(5) We looked at different nuances to check for any patterns. We find some indications that bureaucracy is related to growth in the very short term (but no long-term relationship), and that the relation is stronger in recent decades. But, none of these findings are very robust.

(6) Some general thoughts on null-finding papers in poli.sci.: One curious feature of poli. sci. is the many plausible, widely believed theories that lack strong empirical support. With an explosion of data (of differentt kinds) in recent years, we have our work cut out:

(7) While it is tempting and rewarding to construct new theories, testing old and established ones with fresh evidence is an important (and undervalued) contribution. Having people question long-held assumptions is the first step in moving on and coming up with new knowledge.

(8) A common complaint is that null-results are hard to publish. Generally true, I think, but increased awareness of file-drawer bias, etc., may be pushing our field in the right direction. My experience is that some editors and reviewers really like null-findings.

Big thanks to the @cps_journal editors, @benwansell and @Samuels_DavidJ , for publishing our null results!