Historical Call Reports

Overview

The Office of the Comptroller of the Currency (OCC) was established by the 1863 National Bank Act as the federal regulator of the (at the time) newly established national banks. As part of its supervisory duties, the OCC required national banks to submit periodic statements of condition (balance sheets) reporting their assets, liabilities, and capital. Each year, the OCC compiled these statements into a public report to Congress: the Annual Report of the Comptroller of the Currency.

As part of our research (2022, 2025, 2026), we digitized these statements into a bank-level dataset comprising the balance sheets of all national banks from 1863 to 1941. This data spans the entirety of three important eras in U.S. banking history: the National Banking Era (1863–1913), the Early Federal Reserve (1914-1928), and the Great Depression (1929-1939). See Appendix A of "Failing Banks" for more information about these eras and our dataset (published version; preprint). It includes 370,977 observations across 14,286 unique banks in 53 states and territories.

Note that we continue to periodically update this dataset as we identify typos or other issues and correct them (definitely reach out if you spot any issues!)

Source

The data was digitized from the Annual Report of the Comptroller of the Currency (1863–1941), often using multiple digital versions per year, plus hard-copy editions. The original reports are, for the most part, currently available digitally through FRASER, the Federal Reserve Bank of St. Louis's archival system:

FRASER — Annual Report of the Comptroller of the Currency

There are two exceptions to this:

1. The 1864 Annual Report available on FRASER does not include bank-level balance sheet tables. We have uploaded a scanned version here, based on the documents graciously shared by Peter Huntoon. Peter was also kind enough to donate his carefully assembled physical copies of all 1863–1935 Annual Reports (including supplements), which we use to improve and validate our data. See The Huntoon-Raymond Set for his account of how this collection came together.
2. For 1924 onward, balance sheet tables were published in separate companion volumes (available via Google Books). The relevant table numbers are:

   Year	Table
   1923	Table 89
   1924	Table 93
   1925	Table 94
   1926	Table 98
   1927	Table 121
   1928	Table 97
   1929	Table 103
   1930	Table H
   1931	Table I
   1932	Table J
   1933	Table K
   1934	Table L
   1935	Table M
   1936	Table N
   1937	Table O
   1938	Table P
   1939	Table Q
   1940	Table R
   1941	Table S

Balance Sheet Layout Changes

The format and content of the OCC balance sheets evolved significantly over time. The line items reported in the printed tables were not fixed — they changed as banking practices, regulations, and government financial instruments evolved. There are five main types of changes:

• Name changes: For example, "United States deposits" in 1867 becomes "US deposits" in 1868.
• Splits: A single item could be broken into sub-components. "Due from national banks" (1868) was split into "Due from redeeming agents" and "Due from other banks and bankers" in 1869.
• Combinations: Previously separate items could be merged. "Due from other national banks" and "Due from state banks and bankers" (1873) were combined into "Due from other banks and bankers" (1874).
• Appearances and disappearances: Items appear or disappear due to regulatory changes or changes in government debt instruments. Examples include three percent certificates (1868–1872) and the redemption fund with the US Treasurer.
• Unusual labels: Mostly in later years, including seized collateral (livestock, cotton, etc.) and foreign currency items.

We have standardized variable names when changes were purely cosmetic. However, because information is often gained or lost when items are split or combined (e.g., deposits being split into time and demand deposits), users should familiarize themselves with the meaning of all variables before starting a research project.

For example, between 1863 and 1904, bank deposits corresponded to three categories: deposits (Individual deposits), us_deposits (United States deposits), and usdo_deposits (Deposits of U.S. disbursing officers). In 1905 (a year when Call Reports were particularly terse), only a single variable, "All deposits", was reported, which we assign to deposits. Then, in 1906 this is opened back into us_deposits (United States deposits) and deposits (Individual deposits). This lasts until 1915, when the split was changed into demand_deposits (Demand deposits, including U.S. deposits) and time_deposits (Time deposits). Lastly, for 1929 onwards, reports went back into a single "Total deposits" category, assigned to the variable deposits. Thus, if you wanted to create a single "Total deposits" variable across the entire 1863-1941 period, you should do something like egen total_deposits = rowtotal(deposits us_deposits usdo_deposits individual_deposits demand_deposits time_deposits). If you collapse this series by year, results look like this:

As we can see, this series is smooth except for the year 1905, when part of the "due to other banks" accounts were added into "All deposits" (open question: if any reader knows if this was only deposits of other banks, or something else, do reach out to us!).

Beyond just deposits, please see the Variable Availability Explorer we constructed for an interactive view of which variables are populated in each year, and also our underlying source documentation (layout PDFs and YAML files) for details on the original printed forms.

Variable Definitions

Assets

Liabilities & Capital

Identification & Metadata

Data Quality & Validation

Several validation variables are included in the dataset:

• approx_ok_bs — Balance sheet passes basic validation: asset categories add up within 1% of total assets; same for liabilities.
• ok_bs — As above, but the difference must be under $10.
• is_ambiguous — Very rarely, two banks with the same name operated in the same year, and the charter number match was ambiguous. Affects only 42 observations (21 pairs) out of ~380K (0.01%).
• bs_merge — Indicates whether the match between the Historical NIC panel and the balance sheet was successful. Equals 3 for successful matches, and 1 for observations found only in the Historical NIC panel. The latter occurs in about 1% of the dataset (3,665 obs.), most of which correspond to 1933 onwards.

Version History

This dataset is updated regularly as issues are found and corrected.

Call Report Data

Historical NIC Dependency

The dataset also depends on the Historical NIC panel for bank identification. Changes to the Historical NIC affect the call report data:

Citation

For 1863–1866 or 1905–1941 data

Please cite:

BibTeX:

@article{CorreiaLuckVerner2025,
  title={Failing Banks},
  author={Correia, Sergio and Luck, Stephan and Verner, Emil},
  journal={The Quarterly Journal of Economics},
  year={2025},
  doi={10.1093/qje/qjaf044}
}

For 1867–1904 data

Please cite:

BibTeX:

@article{CarlsonCorreiaLuck2022,
  title={The Effects of Banking Competition on Growth and Financial Stability: Evidence from the National Banking Era},
  author={Carlson, Mark and Correia, Sergio and Luck, Stephan},
  journal={Journal of Political Economy},
  volume={130},
  number={2},
  pages={462--520},
  year={2022}
}

Browse the Data