Danielle,
That's a great design. Maternity leave seems like a policy that should be randomizable - adopting a cluster randomized trial sounds feasible. It would be expensive but since it might improve return to work for women on maternity leave, it might not be extraordinarily expensive for the companies.
Recent work has tried to evaluate the long term health effects for women of being eligible for a generous maternity leave. Because short maternity leave often results in women leaving the labor force or losing their employment, it can have enduring socioeconomic and thus health effects. You might see Avendano and Berkman's recent work (2015) work on this - which I think did use a cross-national comparison.
Maria