I have a cross-section of children from a survey sample (IHDS-2011). I want to cluster the standard errors at the PSU level but there are around 2,252 clusters corresponding to 11,700 observations. The mean size of a cluster is 8.54 (median size is 7). Do I have too many clusters?