In a recent post we compared the average credit score at origination history from NSMO with that from the agency loan-level datasets on an apples-to-apples basis.
In today’s blog, we improved the matching by using the Analysis Weights that are provided in the NSMO dataset to compute the averages. NSMO Analysis Weights are the product of the sampling weight and a non-response adjustment, which can make the NSMO dataset more representative.
After applying NSMO Analysis Weights, we re-ran the query and got the following enhanced charts. We can see the analysis weights bring averages for GSEs and FHA loans closer to the loan-level data compared with those in the last post. However, there is a larger divergence for VA loans. Perhaps it is due to the smaller sample size or coverage bias since the NSMO dataset fails to include all the targeted population, which can make the selected samples unrepresentative.
We checked the number of FHA and VA samples which are 3,401 and 2,487 respectively. It shows the VA sample is not much less than FHA’s. Further research is needed on the large divergence for VA loans between NSMO and the agency loan-level datasets. Researchers can use the NSMO dataset with confidence for GSE and FHA loans, but should use caution with respect to the VA program.