Data Dictionary
The FRBNY Consumer Credit Panel consists of detailed Equifax credit-report data for a unique longitudinal quarterly panel
of individuals and households from 1999 to 2017
1
. The panel is a nationally representative 5% random sample of all
individuals with a social security number and a credit report (usually aged 19 and over). We also sampled all other
individuals living at the same address as the primary sample members, allowing us to track household-level credit and debt
for a random sample of US households. The resulting database includes approximately 44 million individuals in each
quarter. More details regarding the sample design can be found in Lee and van der Klaauw (2010).
2
A comprehensive
overview of the specific content of consumer credit reports is provided in Avery, Calem, Canner and Bostic (2003).
3
The credit
report data in our panel primarily includes information on accounts that have been reported by the creditor within
3 months of the date that the credit records were drawn each quarter. Thus, accounts that are not currently reported on are
excluded. Such accounts may be closed accounts with zero balances, dormant or inactive accounts with no balance, or
accounts that when last reported had a positive balance. The latter accounts include accounts that were either subsequently
sold, transferred, or paid off as well as accounts, particularly derogatory accounts, that are still outstanding but on which the
lender has ceased reporting. According to Avery et al (2003), the latter group of noncurrently reporting accounts, with
positive balances when last reported, accounted for approximately 8% of all credit accounts in their sample. For the vast
majority of these accounts, and particularly for mortgage and installment loans, additional analysis suggested they had been
closed (with zero balance) or transferred.
4
Our exclusion of the latter accounts is comparable to some ‘stale account rules’
used by credit reporting companies, which treat noncurrently reporting revolving and nonrevolving accounts with positive
balances as closed and with zero balance.
All figures shown in the tables and graphs
are based on the 5% random sample of individuals. To reduce processing costs, we
drew a 2% random subsample of these individuals, meaning that the results presented here are for a 0.1% random sample of
individuals with credit reports, or approximately 267,000 individuals as of Q1 2017.
5
In computing several of these statistics,
account was taken of the joint or individual nature of various loan accounts. For example, to minimize biases due to double
counting, in computing individual-level total balances, 50% of the balance associated with each joint account was attributed
to that individual. Per-capita figures are computed by dividing totals for our sample by the total number of people in our
sample, so these figures apply to the population of individuals who have a credit report.
In compari
ng aggregate measures of household debt presented in this report to those included in the Board of Governor’s
Flow Of Funds (FoF) Accounts, there are several important considerations. First, among the different components included in
the FoF household debt measure (which also includes debt of nonprofit organizations), our measures are directly comparable
to two of its components: home mortgage debt and consumer credit. Total mortgage debt and non-mortgage debt in the third
quarter of 2009 were respectively $9.7 and $2.6 trillion, while the comparable amounts in the FoF for the same quarter were
1
Note that reported aggregates, especially in 2003-2004, may reflect some delays in the reporting of student loans by servicers to credit
bureaus which could lead to some undercounting of student loan balances. Quarterly data prior to Q1 2003, excluding student loans, will
remain available on the Household Credit webpage.
2
Lee, D. and W. van der Klaauw, “An introduction to the FRBNY Consumer Credit Panel”, [2010].
3
Avery, R.B., P.S. Calem, G.B. Canner and R.W. Bostic, “An Overview of Consumer Data and Credit Reporting”, Federal Reserve
Bulletin, Feb. 2003, pp 47-73.
4
Avery et al (2003) found that for many nonreported mortgage accounts a new mortgage account appeared around the time the account
stopped being reported, suggesting a refinance or that the servicing was sold. Most revolving and open non-revolving accounts with a
positive balance require monthly payments if they remain open, suggesting the accounts had been closed. Noncurrently reporting
derogatory accounts can remain unchanged and not requiring updating for a long time when the borrower has stopped paying and the
creditor may have stopped trying to collect on the account. Avery et al report that some of these accounts appeared to have been paid off. 5
Due to relatively low occurrence rates we used the full 5% sample for the computation of new foreclosure and bankruptcy rates.
Additionally, to capture and account for servicer discrepancies, we used the 1% sample for student loan data. For all other graphs, we
found the 0.1% sample to provide a very close representation of the 5% sample.