FinObservatory

Deposit market structure / Methodology

How the deposits layer is built

The survey

The FDIC Summary of Deposits (SOD) is, in the FDIC’s words,

“the annual survey of branch office deposits as of June 30 for all FDIC-insured institutions, including insured U.S. branches of foreign banks.”

Every figure in this module is therefore a June 30 snapshot of the survey year, not a year-end or an average. The layer holds the full published span, survey years 1994 through 2025, retrieved from the FDIC BankFind Suite API (api.fdic.gov/banks/sod) as verbatim paginated responses whose per-year row counts match the API’s own totals exactly (retrieval and verification record: docs/data_provenance.md, data/raw/fdic_sod/). FDIC data are a US government work in the public domain (17 U.S.C. 105).

The dollar unit throughout is thousands of US dollars, per the survey’s reporting instructions (Section 1.E): “Report all amounts in thousands, rounded to the nearest thousand dollars.” Pages display these values as $B and $T; no other unit conversion is applied and no series is inflation-adjusted.

The DEPDOM reconciliation anchor

SOD branch deposits are not free-floating: the survey’s reporting instructions (June 2026 edition, Section 2, STEP 3) require each institution to

“Validate that total branch deposits equal total domestic deposits reported on your June 30 ... regulatory financial report ... total branch deposits reported on the Summary of Deposits Survey must equal ‘Deposits in domestic offices’ reported on Schedule RC, item 13.a, of the June 30 ... Call Report.”

That identity, branch sum equals Call Report domestic deposits (DEPDOM), is the layer’s acceptance anchor. Recomputed from the parquet at build time: JPMorgan Chase Bank NA (FDIC cert 628), survey year 2024, sum of deposits over its 4,881 branch rows = 2,008,751,000 thousand USD. The acquisition audit (2026-07-11, docs/data_provenance.md) verified this exact figure against the institution’s DEPDOM on the live FDIC financials API; its total deposits including foreign offices (DEP) differ, as expected, which is why national totals in this module are domestic-office deposits only.

Aggregation

The module reads two derived tables. sod_branches carries the full-fidelity branch rows (one row per branch per survey year). sod_market_structure aggregates them per (year, state) and per (year, state, county): branch count, total deposits, institution count, and deposit HHI. The state key is the branch’s state (STALPBR), so an institution’s deposits are credited to the states where its offices sit, not to its charter state. National totals are the sum of the state rows; the national institution count is the number of distinct FDIC certificates in the branch file that year, because summing per-state institution counts would count a multi-state bank once per state. Unit banks (a single office) enter from their Call Report totals per the survey’s instructions.

The HHI

Concentration is measured with the Herfindahl-Hirschman index over institutions within a geography (a state or a county). With di the summed branch deposits of institution i inside the geography and D the geography’s total, each share is si= 100 × di / D and

HHI = Σi si2,   0 ≤ HHI ≤ 10,000

A geography served by a single institution scores exactly 10,000. The institution is the FDIC certificate: separately chartered banks under one holding company count separately.

The concentration bands drawn on the state pages follow the federal antitrust agencies. The 2010 Horizontal Merger Guidelines (section 5.3) classify markets as unconcentrated below HHI 1,500, moderately concentrated between 1,500 and 2,500, and highly concentrated above 2,500. The 2023 Merger Guidelines (Guideline 1) tightened the screen: markets with an HHI above 1,800 are highly concentrated, and a merger producing a post-merger HHI above 1,800 with a change of more than 100 points is presumed to substantially lessen competition. The 1,500/2,500 bands shown are the 2010 classification; the dashed 1,800 line is the 2023 threshold. Note the attribution: 1,500 and 2,500 come from the 2010 guidelines, 1,800 from the 2023 guidelines.

A statewide or countywide HHI is a descriptive statistic, not a merger-review market share. Banking merger analysis defines geographic markets locally (metropolitan areas and rural counties, with deposit-weighted overlaps), so the figures here are not comparable to the HHIs the Federal Reserve or DOJ compute in a specific bank-merger case.

Caveats

  • Deposits are office-assigned. The survey requires the branch sum to reconcile to DEPDOM, but how an institution allocates deposits across its own offices reflects its internal record-keeping practices, and those practices vary. Deposits without a natural branch home (internet, brokered, sweep balances) are often booked to a main office, which can swell the state hosting a headquarters. Cross-state comparisons should be read with this in mind.
  • Territories are included. The state file covers the 50 states, DC, and the reporting territories and freely-associated states (PR, GU, VI, AS, MP, FM, MH, PW). Several of the Pacific jurisdictions host only one to three institutions, so their HHIs sit at or near the 10,000 ceiling by construction; the index page labels them non-state. One jurisdiction’s coverage ends before the latest survey year; its page carries a dated coverage note.
  • June 30 only. One observation per year. Intra-year swings (for example the March 2023 deposit runs) are invisible between consecutive June snapshots.
  • Domestic offices only. Foreign-office deposits of US banks are outside the survey, so totals here are below measures that include them.
  • Nominal dollars. No deflation or per-capita scaling; long-run growth mixes real growth and inflation.

Sources