What is the HCAI?
The Housing Finance Policy Center’s credit availability index, or HCAI, measures the difficulty of getting a mortgage in the United States by precisely quantifying lenders’ tolerance for risk. Uniquely and importantly, the HCAI separates borrower risk from product risk.
The HCAI calculates the percentage of owner-occupied purchase loans that are likely to default—that is, go unpaid for more than 90 days past their due date. The methodology explicitly incorporates weights for the likelihood of economic downturns. It is an objective, transparent, and precise measurement of mortgage credit availability.
What does the HCAI rating mean?
A lower HCAI indicates that lenders are unwilling to tolerate defaults and are imposing tighter lending standards, making it harder to get a loan. A higher HCAI indicates that lenders are willing to tolerate defaults and are taking more risks, making it easier to get a loan.
How long has the HCAI been in use?
The index was first introduced in November 2014, covering the mortgage market from the first quarter of 1998 through the fourth quarter of 2013. An update was issued in March 2015 and has been issued quarterly since then. The methodology used to calculate the HCAI was updated in October, 2020 to allow for a more accurate calibration when market shares change.
Has the methodology been updated since 2014?
Yes. We updated the methodology in October 2020 to more accurately reflect credit availability changes when there are sharp changes in market shares, by using a 4-quarter moving average rather than relying solely on the last quarter of data. We recalculated all of the quarterly HCAI numbers at that time to reflect these changes. You can read more about the methodology update here. Any HCAI data obtained before October 2020 is not comparable to the current, revised data and should be replaced with the new data.
How does the HCAI work?
To determine the default risk of mortgage loans, the HCAI probes an enormous database of historical loans to find the loans that most closely match the characteristics of the new loans being studied. The characteristics used to match the new loans with the historical loans include borrower characteristics (credit score, loan-to-value, debt-to-income ratio) and loan risk, separating loans with and without risky features (less than five years to the reset, interest-only loans, negatively amortizing loans).
The HCAI searches for and uses two historical default rates to determine the likely default risk of a newly originated loan:
- Normal-year default rate:The default rate of historical loans with similar characteristics originated in a year followed by normal economic conditions (2001 and 2002). A default is defined as a seriously delinquent (90-plus days) payment, including various stages of foreclosure.
- Stressed-year default rate:The default rate of historical loans with similar characteristics originated in years followed immediately by economic stress (2005 and 2006).
The HCAI weights these two default rates according to the likelihood of normal and stressed scenarios, assigning a 90 percent chance for a normal economic environment and a 10 percent chance for a stressed economic environment. The resulting number is the expected default risk the market takes in making this group of loans.
More details about the HCAI’s methodology are available here and information about the October, 2020 update is available here.
How does HCAI differ from other credit availability indices?
Other commonly used measurements of credit availability are variously narrow, subjective, limited in time, opaque, or inaccurate in certain situations. The HCAI is robust, objective, and intuitive because it weights the likelihood of economic downturns and takes borrower and loan characteristics into account. It is also completely transparent.
How does the HCAI compare to the Federal Reserve Board’s Senior Loan Officer Survey on Lending Standards?
The SLO, usually conducted four times a year by the Federal Reserve’s Board of Governors, is designed to measure credit accessibility qualitatively by asking banks to report changes in their lending practices over the previous three months. Currently, senior loan officers at up to 60 large domestically chartered commercial banks respond to the voluntary SLO survey.
Using the SLO, researchers can calculate the net share of domestic respondents tightening lending standards for residential mortgage loans: the fraction of banks that report having tightened standards “considerably” or “somewhat” minus the fraction of banks that report having eased standards “considerably” or “somewhat” over the past three months. Because the SLO compares the current month against the last three, however, it is unable to offer much sense of the change in credit accessibility over a longer period, such as a year or several years.
How does the HCAI compare with the Mortgage Banker Association’s MCAI?
The monthly credit availability index produced by the Mortgage Bankers Association originated in 2012. AllRegs scans the credit guidelines for a large number of lenders, and the MBA aggregates the results into a single number. Though we know that the MCAI takes many factors into account (such as loan purpose, amortization type, and property type), we have no way of knowing how the MBA assigns numbers to these factors or of assessing the formula that converts these many factors into a single index number.
What does the HCAI tell us?
The HCAI allows us to identify with great specificity the likelihood that a particular loan or group of loans will default and how much of that default risk stems from the loan versus the borrower. Put simply, it answers this question: is this a risky loan made to a low-risk borrower or a low-risk loan made to a risky borrower? The HCAI can answer this question about any group of current or historical loans.
What insights has the HCAI provided?
One of the HCAI’s most important insights is that product risk, not borrower risk, fueled the housing bubble. The HCAI shows that the mortgage default risk the market was willing to take peaked significantly from 2005 to 2008. This finding makes sense: we know risky lending was going on during this period. The index also shows us, however, that in the bubble years, the market took almost twice the product risk it took in the pre-bubble years, while borrower risk held steady.
The HCAI also reveals that after the crisis, the mortgage market almost stopped providing loans with risky terms and significantly curtailed its willingness to accept any borrower risk, dipping well below the level held steadily from 1998 to 2007. More recently, lenders have been taking only a fraction of the borrower risk they consistently maintained in the pre-bubble and bubble years—a tremendous overcorrection.
Where do the HCAI data come from?
The HCAI uses the quarterly first-lien owner-occupied origination volume from Inside Mortgage Finance (IMF). The HCAI further separates the mortgage market into three channels: FVR (FHA, VA and Rural Housing and Development) channel, the government-sponsored enterprises’ channel (GSE), and the portfolio and private-label securitization (PP) channel. The HCAI relies on eMBS Inc.’s loan level data for the mortgages from GSE and FVR channels, and relies on Black Knight’s Mcdash servicing database for the mortgages from the PP channel.
Do you revise historic HCAI numbers?
Since many of our data sources report recent loan originations with delays, we recalculate and, when possible, revise the HCAI for the most recent four quarters each time we make updates. In addition, we updated the methodology in October 2020 to de-emphasize the impact of sharp changes in market shares, and we recalculated all of the quarterly HCAI numbers at that time. You can read more about the methodology update here. Any HCAI data obtained before October 2020 is not comparable to the current, revised data and should be replaced with the new data.
Whom do I contact if I have questions? Can I access the data myself? If so, how?
If you have questions about the HCAI data or methodology, or if you would like to request historical data, please e-mail researcher John Walsh. If you have media-related or general questions about the HCAI, please e-mail communications manager Sheryl Pardo.