FSA Livestock Forage Disaster Program Payments, Derived

This repository estimates how much the USDA Livestock Forage Disaster Program (LFP) would have paid, program years 2008–2025, had three identifiable data issues in its county eligibility determinations not been present:

1. Non-authoritative county boundaries (Type 1). The county cut of the US Drought Monitor is performed by the National Drought Mitigation Center (NDMC) under contract to USDA, on a county boundary dataset the NDMC adopted before the 2008 Farm Bill and has revised only three times since.
2. A percent-of-county coverage threshold (Type 2). The county classification the NDMC reports appears to require a minimum share of the county to be in drought before the class is assigned — a stricter standard than the statute’s trigger on drought “in any area of the county” (7 U.S.C. § 1531(d)(3)).
3. An undocumented cap on eligible payment months (Type 3). The months paid appear to be capped at the whole-month floor of the county’s normal grazing period rather than scaled to drought intensity, a rule that appears in no handbook, notice, statute, or Federal Register rule.

For every Census county, FSA county, program year and pasture type, the archive carries the monthly payments earned under each of four USDM county-aggregation conventions alongside the months FSA actually paid, attributes each discrepancy to at most one of the three issues, and scales each county-year’s actual LFP disbursement by the ratio of months warranted to months paid. Summed across the program, the three issues account for an estimated $5.0 billion–$6.8 billion in foregone payments against $13.8 billion actually disbursed.

📂 View the LFP payments archive listing here.

Note: These are estimates with stated assumptions, not an accounting of money owed. The archive is derived from the project’s eligibility reanalysis, FSA’s FOIA-obtained determinations and the Farm Payment Files; it is not a record of USDA’s determinations or payments, and it is not evidence of any producer’s entitlement. The analysis was first presented in the project briefing What Drifted Eligibility Data Costs (July 2026).

🗂️ Contents

• fsa-lfp-payments-derived.csv — one record per county pair, program year and pasture type: months earned under each aggregation, months FSA paid, the discrepancy type and the payment ratios
• fsa-lfp-payments-derived.parquet — the same records as Parquet
• fsa-lfp-payments-derived-county-years.csv / .parquet — actual and estimated additional disbursements per FSA county, program year and scenario
• fsa-lfp-payments-derived-annual.csv / .parquet — national totals per program year and scenario
• fsa-lfp-payments-derived.json — the three tables restructured for browsers (see Output Data)
• provenance.json — the inputs read (URL, ETag, last-modified, rows) and the headline totals
• qa-report.txt — validation summary, record counts by discrepancy type, and every record the attribution or the county join could not place
• fsa-lfp-payments-derived.R — processing script
• _manifest.txt — flat index of every file in the S3-hosted mirror

☁️ Archive Hosting & Automated Publishing

Every output is mirrored to S3 and served via CloudFront at https://data.sustainable-fsa.com/fsa-lfp-payments-derived/ (browse the archive listing or _manifest.txt for a flat index). The tables are small, so all of them are also committed to this repository: the archive is readable from a git checkout alone and its history is inspectable commit by commit.

Publishing is handled by fsa-lfp-payments-derived.R via the shared R/s3-archive.R helpers, and runs in GitHub Actions (.github/workflows/fsa-lfp-payments-derived.yaml). It is dispatched each week once the upstream eligibility reanalysis has published, with a cron fallback. A precheck compares the ETags of the three upstream inputs against those recorded in the published provenance.json and skips the run when nothing has changed. The workflow authenticates to AWS via GitHub OIDC, re-renders this README from the freshly written tables, and commits the outputs back to git only if they changed.

The months columns can move weekly as the reanalysis extends; the dollar columns move only when the Farm Payment Files archive is refreshed, which happens about once a year.

📥 Input Data

Everything is read over HTTPS from companion archives in this organization. No FOIA workbook or payment file is read directly.

The Farm Payment Files are read with DuckDB straight off the archive’s _manifest.txt, the access pattern documented in that archive’s README. Only the hive partitions for the program years in scope are fetched.

Program years in scope

The years covered are those FSA has determined — currently 2008–2025, taken from the FOIA archive at run time. The reanalysis runs a year ahead, computing the current season weekly; those rows have nothing to be compared against and are held back until the FOIA archive catches up. Payments for the latest program year are still being disbursed when the payment files are compiled, so that year’s dollar estimates are partial.

🧹 Processing Workflow

The processing script fsa-lfp-payments-derived.R:

1. Reads FSA’s determinations, drops fire eligibility (which carries no payment factor), and fixes the program years in scope.
2. Reduces the reanalysis to months: the highest drought factor per Census county, FSA county, aggregation, program year and pasture type.
3. Builds the universe of determinations from the normal grazing periods mapped onto Census counties through the dd22 crosswalk, keeping both county keys (see Two county keys below).
4. Sums LFP disbursements to FSA county and program year.
5. Assembles the records, zero-filling months where an aggregation earned nothing and where FSA listed no eligibility, and attributes each record to at most one issue.
6. Computes payment ratios and, per FSA county-year and scenario, the smallest and largest ratio among the affected records, then scales that county-year’s disbursement by each.
7. Sums to program years per scenario.
8. Validates the result and writes the QA report.
9. Exports the three tables as CSV and Parquet, the browser JSON, and provenance.json, and publishes to S3.

🔎 Attributing discrepancies

Comparing the months earned under three of the aggregations against the months FSA paid isolates each issue, because each leaves a distinct signature. Let C be the months under Census boundaries, F the months under the FSA/NDMC boundary file, R the months under the classes the NDMC reported, and P the months FSA paid (FSA Payment Factor).

The conditions are evaluated in order and a record takes the first that holds. Type 3 records are further split by Discrepancy Subtype:

Grazing-period length is days ÷ 30, the convention the briefing used.

Payment ratios

Discrepancy Payment Ratio is the months warranted under the issue’s own comparison divided by the months paid: C/P for Type 1, F/P for Type 2, R/P for Type 3, exactly 1 for None, and missing for Multiple. Census Payment Ratio is C/P for every record, whatever the type: the all-in comparison that bypasses attribution.

A ratio is missing where either side is zero. With P = 0 there is no disbursement to scale; with the warranted months at zero the ratio would claim a full clawback, which this archive does not estimate. Such records are classified, counted in qa-report.txt, and contribute no dollars. Because they are excluded, the dollar estimates are conservative in one specific way: a county FSA never listed as eligible contributes nothing even where the drought record supports payment.

The comparison is on FSA’s Payment Factor, never its Drought Factor. Payment Factor is the payable figure, min(Drought Factor, Maximum Eligible Payment Months); comparing on it is what makes the cap visible as Type 3.

From months to dollars

The payment files record disbursements by FSA county and program year, with no breakdown by pasture type or Census county. Each affected county-year’s actual disbursement is therefore scaled by the ratio of months warranted to months paid, and because an FSA office carries several pasture types and may cover several Census counties, each with its own ratio, the smallest and largest ratio among the affected records bracket the estimate instead of resolving it to a point:

additional disbursement = (ratio − 1) × actual disbursement

for the minimum and the maximum ratio in turn. Scaling assumes the same producers and acreage would have enrolled at the higher factor, which is the natural reading but not a certainty. All amounts are nominal, unadjusted for inflation.

📤 Output Data

Prefer the Parquet where you can. It carries types, so FIPS and FSA County come back as character and the discrepancy columns as factors. The CSV carries none, and base R’s read.csv() and pandas will read the county codes as integers and drop the leading zero, turning Autauga County, AL (01001) into 1001. Read them as character explicitly:

readr::read_csv(
  "https://data.sustainable-fsa.com/fsa-lfp-payments-derived/fsa-lfp-payments-derived.csv",
  col_types = readr::cols(FIPS = "c", `FSA County` = "c")
)

fsa-lfp-payments-derived.csv and .parquet

One record per Census county, FSA county, program year and pasture type.

A handful of FSA determinations have no published normal grazing period. They are kept with their FSA columns filled, every other value missing, and no discrepancy type; qa-report.txt enumerates them.

fsa-lfp-payments-derived-county-years.csv and .parquet

One row per FSA county, program year and scenario, for county-years with at least one affected record.

The first three scenarios use Discrepancy Payment Ratio on records of that type. All Issues (Census) uses Census Payment Ratio on every record where it differs from 1, whatever the type.

fsa-lfp-payments-derived-annual.csv and .parquet

One row per program year and scenario, every year in scope.

All Issues (Sum of Types) and All Issues (Census) estimate the same quantity two ways — by attribution and by direct recomputation — and should agree closely. Their difference is a check on the attribution.

fsa-lfp-payments-derived.json

The same three tables restructured for direct use in a browser, in the layout of the LFP eligibility archives’ payload: column-oriented (one array per field), dictionary-coded (pasture types, FSA county codes and FIPS codes appear once in lookup arrays; the record arrays hold 0-based indices into them), and sorted on the integer index columns so the file gzips well. It is a display format, not an archive-of-record format; for analysis, use the CSV or Parquet.

The payload is self-describing via its schema field (fsa-lfp-payments-derived/1), a frozen contract: fields may be added, but existing ones are never renamed or reordered without bumping the schema.

Nulls appear only where the tables have them: m_* and disc on records with no published grazing period, m_cap where FSA reported none, sub outside Type 3, r_disc and r_census where a ratio is undefined, and paid, lo, hi where a county-year has no recorded disbursement. Dollars are in cents precision; ratios are exact.

🗺️ Two county keys

An LFP determination needs two counties. The normal grazing period is set per FSA county, the administrative unit; eligibility triggers on drought “in any area of the county” as a Census county, the unit the US Drought Monitor is aggregated to. The two do not nest in either direction — one FSA office may cover several Census counties, and several offices may split one — so the records table carries the pair unreduced, as the eligibility reanalysis does.

Disbursements are recorded per FSA county, so dollars are estimated at FSA county grain. Every Census county an office covers contributes its ratio to that office’s minimum and maximum, and a Census county administered by several offices contributes to each office’s own disbursement separately, so nothing is double counted. The Records column says how many records bracket each county-year.

📍 Quick Start: Chart the estimated shortfall in R

This README is rendered by the weekly build, so the example reads the Parquet file that build just produced. To run it yourself, substitute the published URL https://data.sustainable-fsa.com/fsa-lfp-payments-derived/fsa-lfp-payments-derived-annual.parquet — arrow::read_parquet() takes it directly.

library(dplyr)
library(ggplot2)
library(tidyr)

annual <- arrow::read_parquet("fsa-lfp-payments-derived-annual.parquet") |>
  filter(Scenario == "All Issues (Sum of Types)")

paid <- annual |>
  transmute(`Program Year`,
            Additional = `Minimum Additional Disbursement`,
            Paid = `Disbursement Amount`) |>
  pivot_longer(-`Program Year`, names_to = "part", values_to = "amount") |>
  mutate(part = factor(part, levels = c("Additional", "Paid")))

ggplot() +
  geom_col(data = paid, aes(`Program Year`, amount / 1e9, fill = part),
           width = 0.74) +
  geom_linerange(
    data = annual,
    aes(x = `Program Year`,
        ymin = (`Disbursement Amount` + `Minimum Additional Disbursement`) / 1e9,
        ymax = (`Disbursement Amount` + `Maximum Additional Disbursement`) / 1e9),
    colour = "#D9A441", linewidth = 0.9
  ) +
  geom_point(
    data = annual,
    aes(x = `Program Year`,
        y = (`Disbursement Amount` + `Maximum Additional Disbursement`) / 1e9),
    colour = "#D9A441", size = 1.2
  ) +
  scale_fill_manual(
    values = c(Paid = "#B7410E", Additional = "#fdebbb"),
    labels = c(Paid = "Actually disbursed",
               Additional = "Estimated additional payments (low–high)"),
    breaks = c("Paid", "Additional"), name = NULL
  ) +
  scale_x_continuous(breaks = seq(min(annual$`Program Year`),
                                  max(annual$`Program Year`), 2),
                     minor_breaks = NULL) +
  scale_y_continuous(labels = \(x) ifelse(x == 0, "$0", paste0("$", x, "B")),
                     expand = expansion(mult = c(0, 0.04))) +
  labs(
    x = NULL, y = NULL,
    title = "Estimated additional LFP payments by program year",
    subtitle = "Actual disbursements, with the low–high estimate of payments the drought record and the statute would support"
  ) +
  theme_minimal(base_size = 11) +
  theme(legend.position = "top", legend.justification = "left",
        panel.grid.major.x = element_blank(), panel.grid.minor = element_blank())

🧭 Related archives

• fsa-lfp-eligibility — FSA’s determinations as obtained by FOIA, program years 2008–2025
• fsa-lfp-eligibility-derived — eligibility recomputed from the US Drought Monitor under four county aggregations; the months columns here come from it
• fsa-payment-files — the Farm Payment Files, 2004–present; the dollars here come from it
• fsa-normal-grazing-period — the normal grazing periods that define the universe of determinations

📝 Citation

If you use this data in published work, please cite:

Bocinsky, R. Kyle. Livestock Forage Disaster Program Payments, 2008–2025: Foregone Payments Derived from County Eligibility Discrepancies. Montana Climate Office, University of Montana. Sustainable FSA project. Accessed YYYY-MM-DD. https://data.sustainable-fsa.com/fsa-lfp-payments-derived/

Machine-readable metadata are in CITATION.cff; GitHub’s Cite this repository button (top right of the repo page) renders it as APA or BibTeX.

The underlying data this archive reads should be cited separately — the eligibility reanalysis, the FOIA determinations, the normal grazing periods and the Farm Payment Files each have their own archive and citation.

Acknowledgment: This work is part of the Enhancing Sustainable Disaster Relief in FSA Programs project, supported by the USDA Office of the Chief Economist, Office of Energy and Environmental Policy, and the USDA Climate Hubs.

📄 License

• Raw USDM data (NDMC), raw FOIA data and Farm Payment Files (USDA): Public Domain (17 USC § 105)
• Processed data & scripts: © R. Kyle Bocinsky, released under CC0 and MIT License as applicable

⚠️ Disclaimer

This dataset is archived for research and educational use only. It is an estimate derived from other archives, not a record of USDA’s determinations or payments, and it is not evidence of any producer’s eligibility for, or entitlement under, any program. It may not reflect current USDA administrative boundaries or official LFP policy. Always consult your local FSA office for the latest program guidance.

To locate your nearest USDA Farm Service Agency office, use the USDA Service Center Locator:

🔗 USDA Service Center Locator

👏 Acknowledgment

This project is part of:

Enhancing Sustainable Disaster Relief in FSA Programs Supported by USDA OCE/OEEP and USDA Climate Hubs Prepared by the Montana Climate Office

✉️ Contact

R. Kyle Bocinsky Director of Climate Extension Montana Climate Office 📧 kyle.bocinsky@umontana.edu 🌐 https://climate.umt.edu