Researchers use HMDA data to calculate denial rates as a measurernof credit accessibility, usually as the number ofrnapplications denied divided by the total number of applications.  This fails to take into account thernapplicants’ credit profiles.  An increasernin applications by weaker-credit borrowers could increase the denial rate andrnindicate tighter credit access even when the actual credit environment has notrnchanged.  This accounts for the counterintuitive trendrnin figure 1.B: denial rates increaserneach year from 2002 throughrnthe boom, reaching a peak in 2007 and 2008, and fall again as the bottomrnfalls out of the marketrnin 2009.</p

Li and Goodman say that the MBA’s MCAI whichrnis relatively new, uses AllRegs to scan therncredit guidelines for a largernnumber of lenders, then aggregates the results intorna single number.  The resulting trendrnbetween 2004 and the present, shown in figure 1C is quite reasonablernthey say, but the index lacks transparency. “Though we know,rnfor example, thatrnthis index takesrnmany factors intornaccount (such as loan purpose,rnamortization type, and property type)rnwe  havernno way to know how numbers are assigned to these factorsrnor to assess the formularnthat converts these manyrnfactors into a single indexrnnumber.”</p

The fourth measure of creditrnaccessibility is the credit characteristics of loans made to the medianrnborrower.  This does reflect the amountrnof risk the market is willing and expected to take at a given time, for examplernthe big jump in credit scores after the financial crisis.  But the measure, in the authors’ eyes, hasrntwo weaknesses.  First, a credit scorernalone is insufficient as a measure of credit availability (disregarding as it doesrnLTV, DTI, and other metrics) and the results using this measure arerncounterintuitive.  They show a slightrnincrease in medianrnFICO scores for the private-label and bank portfolio channelrnbetween 2000 and 2006, indicating declining credit accessibility over a periodrnwidely perceived as driven by increasing accessibility.  Similar to what was noted about the HMDArndata, this was driven less by the amount of credit in the market than by thernkinds of people applying for loans and the kinds of loans they were seeking.   </p

Li and Goodman say that conceptually the marketrnincreases credit availability by taking more default risk so an alternative tornthe four existing measures is to look at the amount of default risk the marketrntakes on at a given time. This gives a single comprehensive snapshot of thernmarket that takes into account both accessibility and risk.  Second, this measure more directly tracksrnmarket reality; a lower value indicates the market is taking less risk and makingrncredit less accessible; a high value indicates the reverse.</p

Torndetermine the default risk of a newly originated home purchase loan, operationallyrnthe HCAI probes an enormous database of historic loans to find a set that mostrnclosely matches the borrower characteristics and loan risk of the loan ofrninterest.  It then searches for and uses twornhistoric default rates to determine the likely default risk of a newlyrnoriginated loan:</p<ol start="1" type="1"