This paper studies what induces governments to undertake reforms aimed at financial development. Its starting point is Abiad and Mody (AER 95(1), 2005). Rather than their ordered logit technique, it uses a within groups approach allowing for error dependence across countries and over time. This paper finds that policy change in a country is negatively rather positively associated with its liberalization level, while the regional liberalization gap does not appear relevant. On the effects of shocks and crises, it argues that some of the Abiad and Mody (2005) findings are robust, but others are fragile. Furthermore, it claims that the extent of democracy is important for this analysis, and identifies a negative effect of the extent of democracy on policy reform.