DP15791 Effectiveness of collective action against the pandemic: Is there a difference between democratic and authoritarian regimes?

Author(s): Alex Cukierman
Publication Date: February 2021
Keyword(s): collective action, democracy, lockdowns, pandemic deaths, stringency
JEL(s): C30, D7, I1, J1, P16
Programme Areas: Public Economics
Link to this Page: cepr.org/active/publications/discussion_papers/dp.php?dpno=15791

Due to its high contagiousness, and the lags in development and administration of vaccines, containment of the COVID-19 pandemic is highly dependent on public behavior and on the focus and transparency of instructions issued by governing bodies. Democratic governments can mobilize support for painful measures if their decisions inspire broad-based confidence and legitimacy. Mobilizing public support for well-focused sanitary actions can also be achieved by coercion. Although coercive measures, can be used temporarily by democratic governments authoritarian governments have a comparative advantage in enforcing them. Results from a cross section of over 150 countries show that, in the absence of controls cumulative death per million people (CD) are lower in less democratic countries. When controlling for the fraction of old population and other variables the impact of democracy on CD in the entire sample vanishes. But splitting the sample into high democracy countries and low democracy countries reveals that mobilization of collective action is more (less) effective in the first (second) group the higher the level of democracy. An overtime average of the stringency of government responses to the pandemic (S) has a highly significant positive impact on CD suggesting reverse causality from CD to stringency. The paper formulates this dual relation as a 2x2 simultaneous model with a CD schedule and an S schedule and shows theoretically that the observed intersection points (CD, S) nearly trace a relatively immobile governmental, positively sloped, stringency response schedule. An overtime -- cross country estimate of the S schedule confirms this result yielding a highly significant but small coefficient.