Robustness of Shared Prosperity Estimates: How Different Methodological Choices Matter