Move all MNW algorithms (bootstrap types 11, 13, 31, 33) to sparse matrices - more efficient estimation with high dimensional fixed effects. Additionally, types "x3" will also work with fixed effects that are not nested within clusters. fyi @kylebutts
Move all MNW algorithms (bootstrap types 11, 13, 31, 33) to sparse matrices - more efficient estimation with high dimensional fixed effects. Additionally, types "x3" will also work with fixed effects that are not nested within clusters. fyi @kylebutts