Wednesday, November 13, 2013

CY 2010-CY 2012 Data Trifecta: Census Tract Level GSE SF Loan Data; $50 Billion in Loans to 252,000 Oregon Families.

I have prepared a one of a kind 95MB Excel workbook HERE that contains census tract loan level data for $50 Billion/ 252,000+ Oregon single family loans acquired by the GSE's in Oregon during CY 2012, CY 2011, and CY 2010.(I extracted this data from FHFA public use databases for these years).

The data worksheet has columns I have added to FHFA data to show names instead of codes for several selected fields including borrower and co-borrower race and ethnicity, loan purpose, and borrower income ranges; with 48 fields of data for each record, there are more than 12 million pieces of data available for these 3 years . 

A complete list of all fields in the data downloaded from FHFA (a data dictionary) for 2012 is HERE, 2011 HERE, and 2010 HERE [ Census tract definitions, minority percentages, and median incomes may vary from year to year as newer information became available]. 

Using filters and the pivot table in the Excel workbook users can research answers to questions like these:
1. How many loans were acquired that were made to purchase a home where there was a first time Hispanic Borrower or Co-borrower, and where were those loan made (A similar analysis is possible by race).
2. How many/where were loans acquired for borrowers with incomes below 80% MFI? 
3. How many /where were loans acquired that had been made to investors, and where were those loans? 
4. Who got high cost loans and where were they? 
5. How did refinance loan volume change from year to year?

Note that unlike HMDA data, the GSE data covers ALL areas of the state, not just metro areas, and also includes a first time home buyer field not found in HMDA data.

The PDF file HERE and embedded below summarizes by CY and by county and by year the unpaid loan balance at the time of GSE acquisition and the count of loans.

Originally created and posted on the Oregon Housing Blog

No comments:

Post a Comment