Abstract
Cluster-corrected standard errors are widely used but may sometimes be inappropriate since household surveys are increasingly geo-referenced. Compared with the appropriate spatial error models that use details on exact locations, cluster corrections impose untested restrictions on spatial correlations. Our example from rural Indonesia suggests cluster corrections are too conservative and may cause inference errors.
Original language | English |
---|---|
Pages (from-to) | 1857 - 1863 |
Number of pages | 7 |
Journal | Economics Bulletin |
Volume | 34 |
Issue number | 3 |
Publication status | Published - 2014 |