claritystorm/usda-crop-insurance-indemnities-weather

---
license: cc-by-4.0
task_categories:
- tabular-regression
- tabular-classification
tags:
- agriculture
- crop-insurance
- climate
- drought
- weather
- USDA
- geospatial
- finance
pretty_name: USDA Crop Insurance Indemnities + Weather Anomaly 1989-2023
size_categories:
- 1M<n<10M
---

# USDA Crop Insurance Indemnities + Weather Anomaly 1989–2023

**2M+ county × crop × year × cause-of-loss indemnity records** from the USDA Risk Management Agency (RMA) Cause of Loss database — every federally insured crop loss paid in the US since 1989. Pre-joined with NOAA nClimDiv county-level drought (PDSI), precipitation, and temperature anomalies.

**This repository contains a 1,000-row sample (public domain).**
Full dataset (3 tables, CSV + Parquet) → **[claritystorm.com/datasets/usda-crop-insurance](https://claritystorm.com/datasets/usda-crop-insurance)**

## Quick Start

```python
from datasets import load_dataset
import pandas as pd

# Load the 1,000-row sample
ds = load_dataset("claritystorm/usda-crop-insurance-indemnities-weather")
df = ds["train"].to_pandas()

# Top crops by indemnity paid
print(df.groupby("commodity_name")["indemnity"].sum().sort_values(ascending=False).head(10))

# Drought-driven losses
drought = df[df["cause_category"] == "drought"]
print(f"Drought losses: ${drought['indemnity'].sum():,.0f}")
print(f"Drought as % of total: {len(drought)/len(df)*100:.1f}%")

# Loss ratio by cause category
print(df.groupby("cause_category")["loss_ratio"].mean().sort_values(ascending=False))
```

## Tags

- agriculture, crop insurance, USDA RMA, drought, PDSI, climate risk, farm finance, indemnity

## Schema — Indemnity Table (primary)

| Field | Type | Description |
|-------|------|-------------|
| record_id | int | Unique row identifier |
| commodity_year | int | Crop year (1989–2023) |
| state_fips | string | 2-digit state FIPS code |
| county_fips | string | 5-digit county FIPS code |
| county_name | string | County name |
| state_abbr | string | State abbreviation |
| commodity_code | string | USDA RMA commodity code |
| commodity_name | string | Crop name (Corn, Soybeans, Wheat, etc.) |
| insurance_plan_code | string | Insurance plan code (APH, RP, etc.) |
| insurance_plan_name | string | Insurance plan name |
| coverage_category | string | Coverage category (CAT / BUYUP) |
| cause_of_loss_code | string | RMA cause of loss code |
| cause_of_loss_desc | string | Cause of loss description |
| cause_category | string | Normalized category: drought / flood / hail / frost_freeze / wind / heat / insects_pest / disease / other |
| is_weather_cause | bool | True for weather-driven causes (drought, flood, hail, frost, wind, heat, tornado) |
| month_of_loss | int | Month loss was reported (1–12) |
| policies_indemnified | float | Number of policies that received an indemnity payment |
| unit_months_indemnified | float | Unit months indemnified |
| net_planted_quantity | float | Net planted acres/quantity |
| net_endorsed_quantity | float | Net endorsed acres/quantity |
| liability | float | Total liability (insurance coverage amount) in USD |
| total_premium | float | Total premium collected in USD |
| subsidy | float | Federal premium subsidy in USD |
| indemnity | float | Total indemnity paid in USD |
| loss_ratio | float | Loss ratio (indemnity / total_premium) |

## Schema — Weather Table

| Field | Type | Description |
|-------|------|-------------|
| weather_id | int | Unique row identifier |
| county_fips | string | 5-digit county FIPS (join key) |
| year | int | Year (1895–2023) |
| pdsi_annual_mean | float | Palmer Drought Severity Index — annual mean (<−2 = drought) |
| pdsi_growing_season_mean | float | PDSI — growing season mean (April–September) |
| pcpn_annual_mean | float | Precipitation — annual mean (inches) |
| pcpn_growing_season_mean | float | Precipitation — growing season mean (inches) |
| tmpc_annual_mean | float | Temperature — annual mean (°F) |
| tmpc_growing_season_mean | float | Temperature — growing season mean (°F) |
| drought_class | string | Drought classification: extreme_drought / severe_drought / moderate_drought / mild_drought / near_normal / moist / very_moist |

## Dataset Stats

- **Indemnity records**: 2M+ county × crop × year × cause-of-loss rows
- **Years**: 1989–2023 (35 years)
- **States**: All 50 US states + territories
- **Crops covered**: 130+ commodity types (Corn, Soybeans, Wheat, Cotton, Rice, Sorghum, Tobacco, Vegetables, and more)
- **Weather records**: ~175K county-year rows (PDSI, precipitation, temperature)
- **Source**: USDA RMA + NOAA NCEI nClimDiv

## Top Crops by Historical Indemnity

| Crop | % of Total Indemnity |
|------|---------------------|
| Corn | ~28% |
| Soybeans | ~18% |
| Wheat | ~12% |
| Cotton | ~10% |
| Sorghum | ~5% |
| Other | ~27% |

## Use Cases

- **Climate risk modeling** — quantify how drought/flood anomalies translate to crop insurance losses by county
- **AgTech product development** — county-level crop loss distributions for precision agriculture risk tools
- **Actuarial analysis** — loss ratio trends by crop, cause, and region across 35 years
- **ESG / climate finance** — exposure analysis for agricultural supply chain under climate scenarios
- **Research** — academic study of climate-agriculture vulnerability and federal crop insurance effectiveness

## ⬇️ Get the Full Dataset

| Tier | Price | Includes |
|------|-------|----------|
| Sample | Free | 1,000 rows, public domain (this repo) |
| Complete | $99 | All 3 tables, 2M+ records, CSV + Parquet, commercial license |
| Annual | $199/yr | Complete + annual updates as USDA RMA publishes new data |

👉 **[Purchase at claritystorm.com/datasets/usda-crop-insurance](https://claritystorm.com/datasets/usda-crop-insurance)**

## Source

USDA Risk Management Agency (RMA) — Cause of Loss data.
NOAA NCEI nClimDiv — county climate divisional dataset.
All source data is US federal government work in the **public domain** (17 U.S.C. 105).

Processed and enriched by [ClarityStorm Data](https://claritystorm.com).