Supervisory Priorities Index

Codebook

The distributed file (spi_annual.parquet) is a long-format panel: one row per agency, year, and risk domain.

Fields

FieldDescription
yearCalendar year of the underlying publications
agencyOCC, Fed, or FDIC
domainRisk-domain code (see below)
labelHuman-readable domain name
shareDomain's share of the agency's supervisory attention in that year; shares sum to one within each agency-year

Risk domains

CodeDomainScope
CREDITCredit & LendingCredit risk, lending standards, loan quality, concentration
CAPITALCapitalCapital adequacy, loss reserves, stress testing, prompt corrective action
LIQUIDITYLiquidityLiquidity management, funding stability, deposit behavior, lender of last resort
MARKET_RATEMarket / Rate RiskInterest-rate risk, securities portfolio, derivatives, model risk
GOVERNANCEGovernanceBoard oversight, risk frameworks, internal controls, management quality
OPERATIONSOperations & TechnologyOperational risk, cybersecurity, third-party risk, business continuity, fraud
COMPLIANCEComplianceLegal and regulatory compliance, sanctions, insider conflicts
BSA_AMLBSA / AMLBank Secrecy Act, anti-money-laundering, suspicious-activity monitoring
CONSUMERConsumer ProtectionConsumer compliance, fair lending, community reinvestment
EARNINGSEarningsProfitability, revenue mix, sustainability of earnings
STABILITYSystemic & MacroSystemic risk, macroprudential concerns, cross-border exposure
STRUCTUREStructure & CharteringChartering, branching, mergers and acquisitions, holding-company structure
SUPERVISIONSupervision ProcessExamination procedures, ratings, supervisory process itself
TRUST_FIDUCIARYTrust & FiduciaryTrust powers, fiduciary activities, asset management
PAYMENTSPaymentsPayment systems, settlement, money transmission

Construction

Domain scores are produced by a dictionary method applied to the full text of agency publications, with emphasis weighting validated against independent expert annotation. Construction and validation are described in the accompanying research paper.