Methodology — sponsor_bank_ppp_pnl.csv: Source of Record, FDIC BankFind Suite Financial API

Summary

A methodology note for sponsor_bank_ppp_pnl.csv states that the data come from the FDIC BankFind Suite financial API, a republication of quarterly Call Report data. It covers three sponsor banks, Cross River Bank, Celtic Bank and Customers Bank, across 13 quarters from 2019Q4 to 2022Q4. It cautions that the income fields are year-to-date within each calendar year and gives Cross River FY2019 $42.8M to FY2021 $372.0M and Customers FY2019 $83.4M to FY2021 $321.8M. It also states that PPP fee income is not a standalone line in the Call Report and that Customers Bancorp's 10-Ks disclose $65.5M for FY2020 and $279.2M for FY2021.

Summary drafted by a model from the document's text below and checked by script against that text before publication. It is a navigation aid, not a reading of what the document proves. Where AI is used

Full text

# Methodology — sponsor_bank_ppp_pnl.csv

**Source of record.** FDIC BankFind Suite financial API (`banks.data.fdic.gov/api/financials`),
the FDIC's machine-readable republication of quarterly FFIEC Call Report (FFIEC 031/041)
data. The assignment permits "FFIEC CDR bulk data or FDIC BankFind API"; the API was
used because it exposes the PPP-specific call-report memoranda directly and is
join-ready. The underlying primary filings are the banks' Call Reports, available at
the FFIEC Central Data Repository (cdr.ffiec.gov) by CERT/RSSD if a true source PDF is
later required.

**Population.** Three sponsor banks × 13 quarters, 2019Q4 (pre-PPP baseline) → 2022Q4:
- Cross River Bank — FDIC CERT 58410
- Celtic Bank — FDIC CERT 57056
- Customers Bank — FDIC CERT 34444

**Query.** One call per bank, `filters=CERT:<cert> AND REPDTE:[20191231 TO 20221231]`.
Raw JSON responses are archived in `fdic-api-raw/`; the cleaned long table is
`analysis/sponsor_bank_ppp_pnl.csv`.

## Field definitions (all dollar fields in **$thousands**, FDIC convention)
| CSV column | FDIC field | Meaning |
|---|---|---|
| total_assets_000 | ASSET | Total assets |
| equity_000 | EQ | Total equity capital |
| deposits_000 | DEP | Total deposits |
| net_loans_leases_000 | LNLSNET | Net loans and leases |
| total_interest_income_000 | INTINC | Total interest income (**includes amortized PPP fees**) |
| noninterest_income_000 | NONII | Total noninterest income |
| net_income_000 | NETINC | Net income (quarterly, as reported) |
| yield_on_earning_assets_pct | INTINCY | Yield on earning assets (%) |
| net_interest_margin_pct | NIMY | Net interest margin (%) |
| roa_pct / roe_pct | ROA / ROE | Return on assets / equity (%) |
| provision_loan_loss_000 | ELNATR | Provision for loan/lease losses |
| ppp_loans_num | PPPLNNUM | Number of PPP loans (Call Report Sch. RC-M) |
| ppp_loans_balance_000 | PPPLNBAL | Outstanding balance of PPP loans |
| ppp_loans_pledged_to_ppplf_000 | PPPLNPLG | PPP loans pledged to the PPPLF |

## Caveats (editorial-stance relevant)
1. **No standalone PPP fee line.** SBA PPP processing fees are recognized as an
   adjustment to yield and amortized into **interest income (INTINC)** over the loan's
   life (accelerating on forgiveness/prepayment). The PPP-fee contribution is therefore
   *embedded*, not separately disclosed. Net-income swings are presented as
   PPP-*associated*, evidenced by the timing alignment with the PPP balance run-off —
   not as a claim that 100% of the delta is PPP fees. This is a verified-original-document
   inference, consistent with the project's negative-inference-against-lenders stance:
   the surge is shown from the banks' own regulatory filings, not from PR.
2. **Bank-level, not PPP-level.** All non-PPP-prefixed fields are whole-bank totals.
3. **PPPLNNUM can exceed a bank's own originations** where the bank served as the
   balance-sheet/sponsor lender for fintech partners (e.g. Cross River's count spikes
   to 554,708 in 2021Q2). Treat counts as "PPP loans on/through the bank," not
   "loans the bank itself marketed."
4. **⚠ Income fields are YEAR-TO-DATE, not quarterly.** FDIC/FFIEC income-statement fields
   (`total_interest_income`, `noninterest_income`, `net_income`, `provision_loan_loss`) are
   **cumulative within each calendar year** and reset each Q1 (verified: values rise Q1→Q4
   monotonically; the one exception, Cross River 2022, is a genuine Q4 net loss). Therefore a
   "Q4" income value **is the full-year figure**, and the right comparison is annual:
   Cross River **FY2019 $42.8M → FY2021 $372.0M (≈8.7×)**; Customers **FY2019 $83.4M →
   FY2021 $321.8M (≈3.9×)**. To get a single quarter, difference consecutive YTD values.
   **Balance-sheet** fields (assets, equity, deposits, net loans, `ppp_loans_balance`,
   `ppp_loans_pledged_to_ppplf`) are point-in-time and need no adjustment. 2019Q4 = clean
   pre-program (full-year 2019) baseline.
5. **Disclosed PPP fee figure (Customers only).** The PPP fee income that the call report
   cannot isolate is disclosed in Customers Bancorp's 10-Ks: $65.5M (FY2020), $279.2M (FY2021)
   of PPP interest income incl. origination fees. See
   `analysis/customers_bancorp_ppp_income_disclosed.csv`. Cross River and Celtic are private —
   no equivalent disclosure exists.

File and source

File

METHODOLOGY.md.txt

Size

4,336 bytes

SHA-256

acd060fe06b6ec639925a5272a5589aa441764079c52b8d98423e5b44556f40c

Our copy

METHODOLOGY.md.txt

Original

No public link identified.