We also assess the sensitivity of the results to the fixed-effects structure of our empirical
models. In particular, the within-country specification severely restricts the variation that is used
to estimate the effects of IFRS adoption and enforcement changes, and risks “throwing out the
baby with the bathwater.” We therefore explore several alternative fixed-effect structures that
replace the (demanding) country-quarter fixed effects, but still tighten the specification relative
to the three-trend base model: first, to control for firm-specific time-invariant variables, we
replace country- and industry-fixed effects with firm-fixed effects.36 Second, since the liquidity
models might vary across countries (and over time), we interact the firm-level control variables
with country fixed effects (e.g., country*size). Third, we add separate quarter-year fixed effects
for developed markets to allow for the possibility that emerging and developed economies
exhibit different liquidity trends. Fourth, we include separate size coefficients in each quarter
(size*year-quarter) to accommodate liquidity shocks that affect large firms differently than small
firms. Finally, we address the concern about overstated t-statistics that can arise in difference-indifferences
regressions with a large number of observations from the same firm (Bertrand et al.,
2004). For each firm, we collapse the time-series into a single observation pre- and post-IFRS
by computing means. None of these variations (not tabulated) materially alter the results, and
the inferences remain unchanged