View: Household Credit - Federal Reserve Bank of New York
Household Credit - Federal Reserve Bank of New York
Click link for lots of charts on debt, bankruptcies etc.
All calculations use individual nonprime mortgage loan information obtained from First American CoreLogic's Loan Performance (LP) data set1. The data set provides origination characteristics and monthly loan-level information on securitized subprime and alt-A first-lien loans as well as on nonprime second-lien loans. LP includes variables such as date of origination, Zip code in which the property is located, details of the mortgage contract, underwriting information, current interest rate, balance, and payment record.
National-level charts are constructed using a 1 percent random sample from LP; district- and state-level charts are constructed using a 2 percent random sample.
The charts are divided into four main categories: serious delinquency, roll rate, loss severity, and projected performance by market and asset class.Serious Delinquency Charts - These charts show delinquency estimates by vintage for each asset class (for example, subprime, alt-A)2. For each vintage, we plot the percentage of origination amount in serious delinquency for each market by the number of months since origination. Roll Rates Charts - Roll rate charts present the transition probabilities of loans by asset class. The borrower's status is categorized as one of nine "states": current, thirty days delinquent, sixty days delinquent, ninety days delinquent, in foreclosure, in bankruptcy, real-estate owned (REO), prepaid in full, or prepaid with loss (which includes short sales and REO liquidations). We calculate transition matrices for each market and asset class. The one-step transition probability is the probability of transitioning from one performance status to another in a single month (for example, the probability of a loan's payment status changing from "current" to "thirty days delinquent").3 Loss Severity Charts - These charts show expected losses thirty-six months from the most current date by market or by market and asset class. Expected losses are calculated using the probability of loss thirty-six months from the most current date; this calculation assumes that the most recent transition matrix and loss severity for each date persists for the next thirty-six months5. The charts illustrate this past projection for each historical remittance date. Projected Performance by Market and Asset Class - These charts show the expected losses through 36 months by market or by market and asset class. Expected losses are computed using the probability of loss through 36 months; this is calculated under the assumption that the most recent transition matrix for each date persists through 36 months and that the most recent loss severity for each date persists through 36 months. These graphs illustrate this past projection for each historical remittance date.
There are several caveats and limitations to keep in mind when interpreting these charts. First, as some have pointed out, securitized subprime mortgages differ from those held in portfolios. The latter are not covered by LP, meaning that our estimates may appear larger than they actually are. Second, there are a few inconsistencies in the payment status field (for example, an observation making a two-step jump in one period, say, from thirty days to ninety days past due in one month); these inconsistencies have the potential to affect the transition matrices that we use to produce the roll rate and expected losses charts. At the same time, holding the transition matrix constant for three years is a conservative approach that does not take into account the potential for improved conditions. Finally, LP provides low coverage of nonprime seconds ; thus, the calculations for this asset class are not a good representation of real conditions.
1LP provides 93 percent coverage of securitized, nonprime transactions (the figure is based on the "LP universe" of potential transactions covered - publicly traded, nonagency, residential; it excludes such structures as net interest margins and collateralized debt obligations). 2Serious delinquency is defined as the sum of all loans that are either ninety or more days past due, in foreclosure, real-estate-owned, in bankruptcy, or prepaid with loss. 3Transition probabilities have the property that, given the present state, the future and past states are independent. Formally, Pr(X(n+1)=xn|X1=x1,X2=x2,?,Xn=xn )=Pr(X(n+1)=x|Xn=xn). Since the payment status space is finite, we can represent the transition probability distribution with a transition matrix P, where all rows of P sum to 1, all its elements are non-negative, and its (i,j)th element is equal to pij=Pr(X(n+1)=j|Xn=i).4, where LSRt signifies the loss severity rate as of time t; ISs and LPs represent net interest and principal losses in period s; Cs is the discount rate for the losses in period s; costs corresponds to foreclosure and liquidation costs as well as depreciation of property value; Bk is the outstanding principal balance at time k.5Expected loss = loss frequency * loss severity, where loss frequency is the number of losses per number of exposures.
No comments yet.