Centering GSE 2020 Oregon Loan Acquisitions on Minority Borrowers and Co-Borrowers.

In my most recent post this week I included a link to a new Excel workbook I created that included loan level detail for 163,000 GSE (Fannie Mae and Freddie Mac] loan acquisitions in Oregon in 2020.

I have now added 11 new worksheets to the GSE Excel file I created earlier HERE . The file is slightly larger at 135 MB, but the link and file name remain the same.

These worksheets focus exclusively on minority loans acquired by the GSE‘s in Oregon in 2020.

First, there is a two page summary of loans acquired for borrower or co-borrowers in four minority racial categories and for Hispanic borrowers or co-borrowers. That PDF file is HERE and the summary section is pasted below.

Then there are five pivot tables and five related data worksheets for Hispanic, American Indian or Alaska Native, Asian, Black, and Native Hawaiian Islander or other Pacific Islander borrower or co-borrowers.

Share of  Oregon GSE Loans: Four Minority Racial Categories and Hispanics 

The data shows that 5.9%/13,358 of all 163,089 loan acquisitions were for borrower or co-borrowers in 4 racial minority categories and 8.2%%/9,558 were for Hispanic borrower or co-borrowers.

For (43,640) home purchase loans those shares were 9.4%/4,088 (four minority racial categories) and 7.6%./3,324 (Hispanics).

And for (18,884) first time homebuyer loans those shares were 11.3%/2,136  (four minority racial categories)  and 9.9%/1,869 (Hispanics).

Many GSE Minority Loans Acquired Were FHA Insured Loans

84% of all loans reported by the GSE‘s for four minority racial categories were FHA insured loans. FHA insured loans were 80% of all Hispanic loans acquired by the GSE’s.

That FHA share drops to about 48% for home purchase loans for the four minority racial categories and 44% for Hispanic home purchase loans. 

And, perhaps because of rules that may have restricted how government guaranteed loans can be counted toward GSE goals, the percentage of FHA insured loans for first time homebuyer loans acquired for the four minority racial categories, and for Hispanics, was reported as ZERO.

Note that the GSE FHA loan counts are NOT all 2020 Oregon FHA loans made to borrowers and Co-borrowers in the four racial categories or to Hispanics, just those loans that were FHA insured and then acquired by GSE’s.

Originally created and posted on the Oregon Housing Blog.

Update: 2020 GSE Oregon Loan Purchases: $49 Billion, 163,000 Loans. First Time Homebuyers? $6 Billion, 19,000 Loans.

Update: I have added 11 worksheets to help focus on 2020 Oregon minority home loan acquisitions by the GSE's. 

New post HERE has the details. 

--------------

For 20 years I have downloaded and analyzed  Oregon data on loans purchased by Fannie Mae and Freddie Mac (The "GSE's). In 2001 I used that data to provide reports on the state of African American and Hispanic homeownership in Oregon. (Short version: GSE's weren't doing much to help first time African American and Hispanic homeowners and increasing homeownership rates requires more lending to first time home buyers). 

A chronological listing of my blog posts with GSE as the keyword is HERE, 

Several years ago, thanks to Will White in Senator Merkley's office, we finally got the GSE regulator (FHFA) to release annual GSE data in a MS Access format to make it easier to extract data for analysis in Excel.  Data is available at the state, metro, county, and census tract level. 

FHFA published the GSE 2020 data late last week. This year FHFA has changed the data posted to require combining two tables in Access in order to get all data fields for each loan acquired. Suffice it to say that it is a multiple step PITA to do so. [It would be much more user friendly to split the data, sorted by state name]. 

New Excel File

Thanks to the weekend rain I hunkered down and produced a 125 MB Excel workbook HERE with all the GSE loan data for 2020 for Oregon. I added several columns to convert codes into NAMES that users could understand. [File is large at about 125 MB's].

The GSE data includes many of the same data fields as the HMDA data, but covers more geography. Notably it does NOT include lender identification but DOES include FIRST TIME HOMEBUYER data that is missing in HMDA data. 

Within the Excel workbook are multiple worksheets listed in the READ me file in the workbook. They include.

• A pivot table which allows users to explore topics of interest, including race, ethnicity, gender, loan to value, financing costs, etc. [See pic pasted below for snapshot of Pivot table loan counts by Oregon metro area].
• A listing of all data fields including those that I added. 
• Two subordinate worksheets that include only 2-4 unit properties AND only first time homebuyer purchase loans.  
• A statewide and a county level summary worksheet with counts and dollars for all loans, all purchase loans, and first time home purchase loans. Both worksheets include calculations of the PER UNIT acquisition balance for 2, 3, and 4 unit properties AND how those average PER unit costs compare to ONE unit homes. (Spoiler alert--they are a LOT lower). 

Some statewide observations:

The statewide summary as a PDF is available HERE and embedded below. 

A total of 163,089 loans were purchased with 166,107 units. Two to four unit properties accounted for 2,168 loans and 5,186 units. 

Two to Four Unit PER UNIT Average Principal Balances at Acquisition Were Substantially Lower than for One Unit Properties.

The average mortgage balance at acquisition for ONE unit loans was $300,016. 

On a PER UNIT basis the average mortgage balance at acquisition for two to four unit properties was $128,074. That's 57%/171,942 LESS PER UNIT than a one unit property. 

For two to four unit home purchase acquisitions (NOT refinances) the PER UNIT difference was even higher at 58%./$191,148. 

Originally created and posted on the Oregon Housing Blog. 

Corrected; First Nationwide HMDA 2020 Database of 260,000+ Two to Four Unit Loan Originations Merges Loan Level Data AND Lender Name.

Correction:

I had transposed PER unit values in "Key Observations" section below. Corrected now to reflect that PER UNIT values were HIGHER in 2 unit property vs 4 unit property. This could mean lower PER UNIT rent for units in 4 unit property vs PER UNIT rents in 2 unit property IF amenities and location are seen as equal. 

---

State and local legislation to make it easier to site two to four unit housing is a welcome development. 

However, I have seen little information about the details of actual two to four unit loans, where they are being made, to whom, and with what underwriting standards.  

To begin to better understand these loans I downloaded 2020 HMDA nationwide data for all two to four unit loan applications that resulted in a loan origination, and added the lender details for each loan. 

Note that this does NOT include all loan applications, ONLY those that resulted in a loan origination in 2020 and was reported via HMDA. 

Data is available at the national, state, metro, county, and census tract level for more than 260,000 loans. While this is a substantial number or loans, it is less than 2% of the nearly 13.9+ million ONE unit loan originations reported in HMDA for 2020. 

The 140 MB Excel workbook I created is HERE. [You likely will need to download the file instead of viewing it in your browser]. 

The first READ ME worksheet lists the 10 worksheets in the Excel file. Worksheets include a pivot table to make navigation simpler and a summary page showing by state all loan originations and PER UNIT values by property unit size (two, three, or four units). Also included is a worksheet listing of all 113 data elements that include race and ethnicity, income, debt to income ratios and loan to value ratios. Another worksheet shows all the codes for  each data element. 

I also include a pivot table and worksheet with only OREGON data. It includes 2 additional data fields, county names and metro names. A MSA focused screenshot of the default Oregon pivot table is pasted below. 

Key Observation

Per UNIT values decrease as property unit size increases. Nationally, a 2 unit property had an average PER UNIT value of $249K vs $159K PER UNIT value for a 4 unit property, a 36% reduction. [Oregon had a more modest 21% reduction from $222K to $175K ].  

IF this translates to rents the PER UNIT rent in a 4 unit property should be lower than the PER UNIT RENT in a 2 unit property, IF amenities and location are seen as equal.  

Oregon HMDA 2020 Two to Four Unit Loan Originations by MSA

Originally created and posted on the Oregon Housing Blog. 

Portland Example: First Time Homebuyer Income Required to Qualify for a FHA Oregon Bond Loan Could Be Reduced by Up to 22% Using FHA Compensating Factors.

There are a wide variety of measures that are used to determine homeownership affordability. Those measures can include down payment assumptions, loan rate assumptions, home price assumptions, and allowable debt to income ratio assumptions. 

However I don't recall seeing any projections of the income required to purchase a home that use compensating factors as part of the debt to income analysis. Compensating factors allow applicants to qualify for larger loans than they would qualify for using standard housing debt and total debt to income ratios of 31% for housing expenses and 43% for all reoccurring expenses.

I'm most familiar with FHA lending so my examples will focus on FHA compensating factors. but  I'm sure that other loan programs also have their versions of compensating factors. 

Pre COVID 2019 HMDA Data Shows That the MAJORITY of Oregon FHA Home Purchase Borrowers Had Debt to Income Ratios of 44% or Higher.

In Oregon my review of 2019 HMDA data shows that 52% of FHA home purchase first lien loans originated had debt to income ratios of 44% or higher. (4,404 FHA home purchase first lien loans where DTI was reported at 44% or higher/7,922 total FHA first lien home purchase loans where DTI was known=52%). 

I also looked at Oregon 2019 HMDA data for FHA first lien home purchase loan originations to Hispanic, Black, and Asian borrowers. Their share of FHA first lien home purchase loans originated where DTI was 44% or higher was 66%, 67%, and 67% respectively.

[The link HERE will download the 2019 HMDA Oregon data for your review and further analysis]. 

FHA Compensating Factor Guidance

FHA's Single Family Policy Handbook 4000.1 includes a matrix of manual underwriting compensating factors that can be used to increase the percentage of income that can be used to qualify for an FHA loan.

The table pasted below (from handbook printed page 335) lists those factors that can be used to increase the amount of income used to qualify an applicant for an FHA loan:

A Portland  First Time Homebuyer Metro Example. 

I took the information from that table and applied it to an example I created for the Portland metro area for first time homebuyers. The PDF table I created is HERE and embedded below.

For the Portland metro area I created four scenarios shown in the PDF; a description of the compensating factors required for each scenario at at the top of each column:  

1. All scenarios assume a credit rating of 580 or more. This is, by itself, a compensating factor.
2. All use the current OHCS non targeted area Portland metro maximum purchase price of $453,000, the 2.25% current "cash advantage" interest rate for the Oregon Bond program, and a 3.5%/$15,855 down payment. All scenarios are fixed rate 30 year loans. Note that these terms are generally available only to first time home buyers. There are some exceptions for veterans and also for purchasers of property in targeted areas.  
3. The one time 1.75%/$7,650 FHA upfront mortgage insurance premium is included in the amount financed and the monthly MIP expense is .85%/$310.
4. Monthly property taxes and insurance and homeowners association expenses are estimated at 1.3%  of the purchase price divided by 12. ($491 per month).
5. Total monthly housing expense is $2,387.
6. Monthly non-housing reoccurring debt is assumed at $300 per month except for the 4th scenario where it is zero.
7. None of the scenarios factor in any boost for energy efficient homes, which could add 2% to the front and back ratios. My read is that these factors may only be available for scenarios 1 and 2.

Observations:

1. Using standard FHA underwriting ratios of 31%/43% in the first scenario the applicant would need $96,800 in annual income.
2. Scenario 2 requires only one compensating factor out of three possibilities and would decrease required income to $81,200 and increase allowable underwriting ratios to 37%/47%.
3. In scenario 3 there is an $75,100 income requirement but the applicant must meet at least two of these three compensating factors. The $75,100 income requirement is 22%/$21,700 less than in the first standard scenario. The underwriting ratios for this scenario are further increased to 40%/50%.
4. In scenario 4 (with only ONE required factor--no monthly recurring debt) the annual income required is also $75,100. using underwriting ratios of 40%/40%.
5. Using HUD's 4 person MFI for the Portland metro area of $96,900, 

Scenario 1 requires an income of $96,800---100% of MFI, 
Scenario 2 requires an income of $81,200--84% of HUD MFI, and 
Scenarios 3 and 4 require an income of $75,100--78% of HUD MFI.

NOTE: Use of a higher down payment would reduce monthly housing costs and reduce the income required to qualify in all scenarios, but poses a challenge for first time home buyers since it would also increase the cash required to close the loan.

Caveats:

1. The guidance cited above applies to manual loan underwriting. I do not know the extent to which these specific factors are modified or incorporated in the automated underwriting systems.  However given the large share of loans with DTI of 44% and higher it seems highly like that FHA approved automated credit scoring systems DO use credit scores and other factors in issuing loan approval decisions for loans with DTI at 44% and above. 
2. Lenders are free to adopt more restrictive standards than found in the handbook.  
3. I used the more restrictive income income limit of the two limits in each scenario rounded up to the next higher $100. 
4. The documentation required for each compensating factor can be found starting on PDF page 351 of the 40001.1 handbook.  
5. I used the latest version of the handbook, which is effective in August; I don't see any substantive changes in the compensating factors from those used in the current Handbook.

Originally created and posted on the Oregon Housing Blog.

Oregon 2019 HMDA Loans Data Includes Worksheet and Pivot Table of Black Home Loans.

On June 24th the FFIEC made 2019 HMDA home loan data available for download HERE.  

Unlike prior year HMDA data products from the CFPB, the FFIEC did NOT make it easy to extract data; many of the data fields use codes instead of names. Their browser tool also only allows the selection of two variables at a time for downloads. 

Nonetheless the 2019 HMDA data is useful and has been expanded to include 98 data fields. (These fields are listed in the Excel workbook I created; see below). Also, the 2019 HMDA data was posted on a much more timely basis than I recall when prior year uploads typically happened in September or October of the following year. 

Because the data is valuable and timely I took the time to download the HMDA loan registry (LAR) data for Oregon. I also added  4 columns with NAMES to substitute for codes: Metro names, county names, a name for the outcome of the loan application (Loan originated, denied etc) and a name for loan purpose. Note that the data that I downloaded does not include lender names; that would require downloading and matching loan data with lender data. 

The LARGE Excel workbook (149 MB's I created is HERE.
It contains these worksheets:

1. All data on 2,469 Black loan applications.
2. A pivot table of Black loan applications whose default view is the number of loan applications by county and their disposition. It also includes a filter that can focus on loan purpose (Purchase, refinance ETC) and loan type (FHA, conventional ETC)
3. All data on 264,983 loan applications.
4. A pivot table of all loan applications. The default view is of the count of loan applications and their disposition broken out by race. It also includes a filter that can focus on loan purpose (Purchase, refinance ETC) and loan type (FHA, conventional ETC)
5. A lookup table that includes MSA, county, loan purpose, and disposition values
6. An error table that includes 982 import errors (3.7 errors per 1,000 loans)
7. A table that lists all the 98 fields in the HMDA data (plus the 4 field names that I added.)

Example: Black Loans in 3 County Portland Metro Area. 
Below is a picture of a data table I created within the Black loan pivot table that shows the disposition of 1,837 Black loan applications from three Portland metro counties: 
990 loans were originated, 
67 were purchased (usually by Fannie and Freddie) and 
the remainder [780] did not result in loans. 

Notes: 

• While I included a separate worksheet for Black loan applications it's simple to use a filter within the all loan application worksheet to extract data for other racial and ethnic groups. Look for the "derived ethnicity" and/or the "derived race" columns. The Pivot table for all loan applications will also allow this focus.
• My prior HMDA related posts can be viewed HERE. 

Originally created and posted on the Oregon Housing Blog.

New Report on Mortgage Lending and Foreclosure Disparities.

From Center for Responsible Lending HERE, data includes several state and metro breakouts.

Originally created and posted on the Oregon Housing Blog.

New Tool to Search 233,000 Oregon HMDA 2010 Housing Loan Applications.

In time for the OHCS Housing Conference this week I have put together an Excel workbook HERE with HMDA data for all Oregon MSA's for 2010 [Database also includes Clark and Skamania counties in Washington state as they are in Portland MSA]. 

This Excel 2007/2010 file includes data on 233,512 Oregon residential loan applications in 2010. (NOTE: File is large at 100+ MB's; that's because it contains 71 fields for each loan application/record -more than 16.5 MILLON pieces of data. 

To download the file do NOT attempt to open file within the browser but instead right click file link and save to your PC; be patient as it will take a few minutes to download. A PC with lots of memory and a higher processor speed is recommended.

This database starts with 46 fields of HMDA  2010 data that I downloaded for each Oregon MSA from FFIEC. I then added 25 fields of data to the right of the HMDA data that substitute NAMES values for CODE values that appear in several HMDA fields. [The code sheet for Loan Application Register data from FFIEC is HERE].

Example: HMDA data uses a minority race code of "3" for African American applicants, I added a NAME race code field that that shows "African American".
 
I ALSO merged the HMDA Oregon MSA data with several fields from the "Reporter" HMDA database to add NAMES to codes used to show which lender reported the data, where the lender was located etc. 

The end result is a database that shows at census tract, county, MSA, and state levels:

• Demographic data for each loan applicant, including income, race, ethnicity, and sex.
• Demographic data for the census tract and metro area for the property.
• The loan type [FHA etc], purpose [purchase/refinance etc], loan amount/spread.
• The outcome of each loan application and reasons for denial.
• The lender who reported the loan application result.

As a tease of what's in the workbook the legal sized PDF HERE [worksheet 3 in the workbook] shows for Oregon only, owner occupied home purchase originations by loan type [FHA etc] by ethnicity and then by race.  The table shows :

• That 61.6% of owner occupied home purchase loan originations for Hispanics were FHA loans, compared to an FHA share of owner occupied home purchase orginations to non Hispanics of 39.8%.

• That FHA had 61.9% of the owner occupied home purchase African American loan originations, but only 41.3% of owner occupied home purchase white loan originations

To go deeper I have provided a PIVOT table in the workbook allowing users to EASILY slice and dice the HMDA data in countless ways.

The Pivot default in the workbook is set for Oregon ONLY to retrieve by borrower race, the number of Owner occupied, Home Purchase Principle Residence Loan Originations by loan program (FHA, VA, Conventional, RD).


A short summary of some of the additional data the Pivot table can reveal: 

Subtotal, OREGON Only Owner Occupied Loan Originations Summary, Excluding Manufactured Housing:

• 23,334 Owner Occupied Principle Residence Home Purchase Loan Originations
• 64,128 Owner Occupied Principle Residence Home Refinance Originations

Subtotal, Oregon Only Manufactured Home Loan Originations

• 1,599 Manufactured Home Loan Originations; of those only 576 were Owner Occupied Principle Residence Home Purchase Loan Originations.

Subtotal, Washington State ONLY(Clark County+Skamania County) [Skamania county loans have a  Washington State Code and a "59" county code or a "N/A" code in the County Name Field].

• Excluding Manufactured Housing, 30,276 Loan Applications. 
• 450 Manufactured Housing Loan Applications.

While these are STATE level examples, similar data can be retrieved at the census tract, county, and MSA levels. (Database doesn't include a city data field, but CT data could be combined with other data sources to arrive at summaries at the City level).

The complete list of worksheets in the workbook : 

READ ME
1 Pivot
2 Oregon MSA HMDA Data 2010
3 OO Home Purchase Race OR
4 Race Ethnic  by County
5. Race Eth Originations Cnty
6. 2010 Reporter

Originally created and posted on the Oregon Housing Blog.

2010 HMDA Data is Out.

2010 HMDA data is out; metro aggregate reports from FFIEC website are HERE.  

A future Federal Reserve Bulletin article HERE has one take on what the data shows; look for national and local news stories in the coming days and weeks.

Originally created and posted on the Oregon Housing Blog.

2009 HMDA Oregon Owner Occupied 1-4 Family Summary by County: Home Purchase Loans Were DOWN (Except for FHA) , Refinances WAY Up.

I took a look at the recent Oregon HMDA 2009 data I recently posted (prior post HERE). 

I compared that data with 2008 HMDA data and did a COUNTY and STATE level PDF table HERE that compares data for 2009 and 2008 using these filters: 

1. Loan Originations
2. Owner Occupied
3. 1- 4 family properties 
4. ALL loan sources, with a FHA subtotal.

(Table is formatted as legal to fit all the columns). 

Some observations:

1. Total volume from all sources was up 50%, but this was because refinances were up by 82%, while home purchase loans from all sources declined by 4%.
2. Total volume for FHA was up 55%, because FHA home purchase volume was UP by 60%, while refinances increased by only 50%.
3. Calculating from data on the chart, FHA share of home purchases increased to 39% in 2009 from 23% in 2008; FHA share of refinances declined to 12% in 2009, down from 14% in 2008; FHA share of Purch+Refi increased marginally from 17.6% in 2008 to 18.3% in 2009.
4. More than 3 out of every four (76%) HMDA loans originated in 2009 in Oregon was a REFINANCE. (88,306/116,610). In 2008, refinances represented only 62% of all loans originated (48,454/77,815).

Originally created and posted on the Oregon Housing Blog.

Oregon Housing Blog Releases Complete/Enhanced CY 2009 Oregon HMDA Data, Two Days After National Release.

The FFIEC released national HMDA data on Monday, Sept 20. Their PR is HERE.

While it usually takes many months for Oregon HMDA analysis to begin to appear, and that analysis is often fragmented, I hope to help others get a jump start by providing a COMPLETE set of CY 2009 Oregon HMDA data in one Excel workbook just two days after the national HMDA data release.

• The workbook that I created has data on more than 300,000 Oregon HMDA loan applications, and 205,000 loans made or purchased during CY 2009. 
• For 18 HMDA data fields I added columns to the right of the HMDA downloaded data that show understandable NAMES for what would otherwise be 5.4+million codes. 
• I include documentation showing the process that I used, so that others can replicate and confirm accuracy of the data.
• I created separate worksheets, AND pivot tables, for (A). Applications (B). Loans originated or purchased. Pivot tables will allow users to slice/dice data in an infinite number of ways to meet their needs. 

Three important caveats:

1. Look at the READ ME worksheet when you first open the workbook. 
2. No warranty about accuracy or completeness of data is expressed or implied.
3. File is HUGE, 188 MB's, and to work with data you will need a PC with substantial processing power, both CPU and memory. IT may take 20 minutes or more to download from my web space, so grab a cup of coffee and be patient; time could be even longer if someone else is also downloading file. Before you open file on your PC, I suggest you close ALL other applications.

You can download the 2009 Oregon Housing Blog HMDA Workbook HERE. (NOTE: This workbook was created in Excel 2007 format.Some users report when they save Excel 2007 files they end up with a compressed .zip file extension. My suggestion is to RIGHT CLICK and save the file to your PC. Then navigate to the file you downloaded and look at its file extension. IF it appears as .ZIP extension, change the .ZIP extension to Excel 2007 extension (.xlsx), and THEN open the file with Excel 2007/2010. 

IF you use this data, I would very much appreciate you letting me know how you have used it, and sharing of links to the data analysis that you have completed. You can email me at houspedx@gmail.com 

Originally created and posted on the Oregon Housing Blog.

Oregon HMDA 2008 Data: Sources for Hispanic Owner Occupied Home Purchase Loan Originations.

I am in the process of digesting HMDA 2008 data released last week which includes for information on more than 250,000 loan applications, including 31,000 owner occupied home purchase loan originations.

I have put together a table HERE which shows owner occupied loan home purchase loan originations for ALL borrowers and also shows a breakout of loans originated where there was either a Hispanic Borrower or Co Borrower.

FHA Had 38% Hispanic Market Share Vs. 24% Market Share for All Borrowers
The table shows that FHA had 24% of ALL owner occupied home purchase loan originations, but 38% of the Hispanic owner occupied home purchase market. That means that FHA had a 61% higher representation in the Hispanic owner occupied home purchase market than it had in the overall owner occupied home purchase market.

The table also shows owner occupied home purchase loan originations for conventional lenders, the Rural Housing Service, and Veterans administration.

Although loan numbers were small, Rural Housing Service programs also had higher representation in the Hispanic owner occupied home purchase market than their overall market share, while both VA and conventional loans were UNDER represented in the Hispanic owner occupied home purchase market.

Originally created and posted on the Oregon Housing Blog.

HMDA Oregon, 2008: Hispanic Owner Occupied Home Purchase Orginations by Lender.

Using HMDA 2008 data, I have created a three page PDF file HERE which shows by lender, the number of loan originations made for owner occupied home purchase loans where the borrower or co- borrower was Hispanic.

Originally created and posted on the Oregon Housing Blog.

HMDA 2008: $7.3 Billion in Oregon Owner Occupant Home Purchase Loan Originations by County and Loan Source.

I have prepared a one page PDF file HERE which shows by Oregon county the number and loan amounts for $7.3 Billion/31,000+ in 2008 owner occupied home purchase loan originations from lenders subject to HMDA disclosure requirements.

Multnomah was the county with the largest volume with 7,215 loans and $1.85 Billion.

Originally created and posted on the Oregon Housing Blog.