PPP Loans

Full schema reference, querying examples, aggregate functions, and data quality notes for the PPP loan dataset.

The ppp_loans table contains ~11.5 million Paycheck Protection Program loan records from the SBA's FOIA dataset.

Source

Property	Value
Source	SBA FOIA Portal
Total records	~11,500,000
Last SBA update	September 30, 2024
Refresh strategy	Static (PPP program has ended)

Sample Record

A complete JSON object returned by the API (with select=*):

{
  "id": 7654321,
  "loan_number": "3856107205",
  "date_approved": "2021-02-15",
  "sba_office_code": "0361",
  "processing_method": "PPS",
  "loan_status": "Paid in Full",
  "loan_status_date": "2022-06-01",
  "term": 60,
  "sba_guaranty_percentage": 100,
  "initial_approval_amount": 187500.00,
  "current_approval_amount": 187500.00,
  "undisbursed_amount": 0,
  "forgiveness_amount": 187500.00,
  "forgiveness_date": "2022-03-15",
  "borrower_name": "TONY'S ITALIAN KITCHEN LLC",
  "borrower_address": "1234 MAIN ST",
  "borrower_city": "AUSTIN",
  "borrower_state": "TX",
  "borrower_zip": "78701",
  "business_type": "Limited  Liability Company(LLC)",
  "naics_code": "722511",
  "franchise_name": null,
  "business_age_description": "Existing or more than 2 years old",
  "jobs_reported": 35,
  "non_profit": null,
  "rural_urban_indicator": "U",
  "hubzone_indicator": "N",
  "lmi_indicator": null,
  "race": "Unanswered",
  "ethnicity": "Unknown/NotStated",
  "gender": "Unanswered",
  "veteran": "Unanswered",
  "cd": "TX-25",
  "project_city": "AUSTIN",
  "project_county_name": "TRAVIS",
  "project_state": "TX",
  "project_zip": "78701",
  "servicing_lender_name": "Wells Fargo Bank, National Association",
  "servicing_lender_location_id": "56267",
  "servicing_lender_address": "101 N PHILLIPS AVE",
  "servicing_lender_city": "SIOUX FALLS",
  "servicing_lender_state": "SD",
  "servicing_lender_zip": "57104",
  "originating_lender": "Wells Fargo Bank, National Association",
  "originating_lender_location_id": "56267",
  "originating_lender_city": "SIOUX FALLS",
  "originating_lender_state": "SD",
  "utilities_proceed": 5000.00,
  "payroll_proceed": 165000.00,
  "mortgage_interest_proceed": 0,
  "rent_proceed": 17500.00,
  "refinance_eidl_proceed": 0,
  "health_care_proceed": 0,
  "debt_interest_proceed": 0
}

The naics_code value 722511 means "Full-Service Restaurants" — see the NAICS Code Reference for the full hierarchy.

Schema

Loan Identification

Column	Type	Description
`id`	bigint	Internal primary key
`loan_number`	text	SBA-assigned loan ID (unique)
`date_approved`	date	Loan approval date
`sba_office_code`	text	SBA district office code
`processing_method`	text	`PPP` (first draw) or `PPS` (second draw)
`loan_status`	text	Current loan status
`loan_status_date`	date	Date of last status change
`term`	integer	Loan term in months
`sba_guaranty_percentage`	numeric	SBA guaranty percentage

Financial Amounts

Column	Type	Description
`initial_approval_amount`	numeric	Originally approved amount
`current_approval_amount`	numeric	Current approved amount
`undisbursed_amount`	numeric	Amount not yet disbursed
`forgiveness_amount`	numeric	Amount forgiven
`forgiveness_date`	date	Date forgiveness was processed

Borrower Information

Column	Type	Description
`borrower_name`	text	Business name
`borrower_address`	text	Street address
`borrower_city`	text	City
`borrower_state`	text	2-letter state code
`borrower_zip`	text	5-digit ZIP code

Business Characteristics

Column	Type	Description
`business_type`	text	Business entity type
`naics_code`	text	6-digit NAICS industry code — hierarchical, use `like` for prefix queries
`franchise_name`	text	Franchise name (if applicable)
`business_age_description`	text	Business age description
`jobs_reported`	integer	Number of jobs reported
`non_profit`	text	`Y` or NULL
`rural_urban_indicator`	text	`R` or `U`
`hubzone_indicator`	text	`Y` or `N`
`lmi_indicator`	text	`Y` or NULL

Demographics (Self-Reported)

Column	Type	Description
`race`	text	Race (high NULL rate)
`ethnicity`	text	Ethnicity (high NULL rate)
`gender`	text	Gender (high NULL rate)
`veteran`	text	Veteran status

Geographic / Congressional

Column	Type	Description
`cd`	text	Congressional district (e.g., `CA-12`)
`project_city`	text	Project location city
`project_county_name`	text	Project county
`project_state`	text	Project state
`project_zip`	text	Project ZIP code

Lender Information

Column	Type	Description
`servicing_lender_name`	text	Servicing lender name
`servicing_lender_location_id`	text	Servicing lender location ID
`servicing_lender_address`	text	Servicing lender address
`servicing_lender_city`	text	Servicing lender city
`servicing_lender_state`	text	Servicing lender state
`servicing_lender_zip`	text	Servicing lender ZIP
`originating_lender`	text	Originating lender name
`originating_lender_location_id`	text	Originating lender location ID
`originating_lender_city`	text	Originating lender city
`originating_lender_state`	text	Originating lender state

Proceeds Breakdown

Column	Type	Description
`utilities_proceed`	numeric	Utilities portion
`payroll_proceed`	numeric	Payroll portion
`mortgage_interest_proceed`	numeric	Mortgage interest portion
`rent_proceed`	numeric	Rent portion
`refinance_eidl_proceed`	numeric	EIDL refinance portion
`health_care_proceed`	numeric	Healthcare portion
`debt_interest_proceed`	numeric	Debt interest portion

Search

Column	Type	Description
`borrower_name_search`	tsvector	Full-text search vector (auto-generated)

Fast-Lookup Columns

These columns are optimized for fast filtering and sorting:

loan_number (unique)
borrower_state
borrower_zip
borrower_city (when combined with borrower_state) — case-insensitive matching
naics_code
initial_approval_amount
forgiveness_amount
jobs_reported
date_approved
processing_method
loan_status
borrower_name_search (full-text search)

At least one of these columns must appear as a filter in table queries. borrower_city requires borrower_state as a co-filter (composite index). Guardrail violations return a 400 response.

Querying Examples

For general query syntax (operators, pagination, sorting), see the Query Syntax Reference.

Find businesses by name in a state

curl "https://newbizbot.ai/api/data/ppp_loans?borrower_name_search=fts.starbucks&borrower_state=eq.WA&limit=5" \
  -H "Authorization: Bearer YOUR_API_KEY"

[
  {
    "borrower_name": "STARBUCKS CORPORATION",
    "borrower_city": "SEATTLE",
    "borrower_state": "WA",
    "initial_approval_amount": 0,
    "naics_code": "722515",
    "jobs_reported": 228769,
    "date_approved": "2020-04-10",
    "loan_status": "Exemption 4"
  }
]

Largest loans in an industry

# Top 5 construction loans in Florida
curl "https://newbizbot.ai/api/data/ppp_loans?borrower_state=eq.FL&naics_code=like.23*&order=initial_approval_amount.desc&limit=5&select=borrower_name,borrower_city,initial_approval_amount,naics_code,jobs_reported" \
  -H "Authorization: Bearer YOUR_API_KEY"

[
  {
    "borrower_name": "AJAX BUILDING COMPANY, LLC",
    "borrower_city": "TAMPA",
    "initial_approval_amount": 10000000.00,
    "naics_code": "236220",
    "jobs_reported": 500
  },
  {
    "borrower_name": "COASTAL CONSTRUCTION GROUP OF SOUTH FLORIDA, INC.",
    "borrower_city": "MIAMI",
    "initial_approval_amount": 10000000.00,
    "naics_code": "236220",
    "jobs_reported": 437
  }
]

Loans by ZIP code

# PPP loans in downtown Manhattan (ZIP 10007)
curl "https://newbizbot.ai/api/data/ppp_loans?borrower_zip=eq.10007&order=initial_approval_amount.desc&limit=5" \
  -H "Authorization: Bearer YOUR_API_KEY"

Second-draw loans only

# Second-draw PPP loans (PPS) in healthcare
curl "https://newbizbot.ai/api/data/ppp_loans?processing_method=eq.PPS&naics_code=like.62*&borrower_state=eq.TX&limit=10" \
  -H "Authorization: Bearer YOUR_API_KEY"

Loans without forgiveness

# Loans in California that haven't been forgiven yet
curl "https://newbizbot.ai/api/data/ppp_loans?borrower_state=eq.CA&forgiveness_amount=is.null&initial_approval_amount=gt.100000&order=initial_approval_amount.desc&limit=10" \
  -H "Authorization: Bearer YOUR_API_KEY"

Non-profit organizations

# Non-profit PPP recipients in New York
curl "https://newbizbot.ai/api/data/ppp_loans?non_profit=eq.Y&borrower_state=eq.NY&order=initial_approval_amount.desc&limit=10&select=borrower_name,borrower_city,initial_approval_amount,naics_code,jobs_reported" \
  -H "Authorization: Bearer YOUR_API_KEY"

Natural Language → API Translation

These examples show how to translate common human questions into API calls. Each example breaks down the reasoning for filter selection.

"Find one-person legal recruiting firms in the DC/MD/VA area"

Thinking: "Legal recruiting" = staffing industry (NAICS 561311, 561312) + "legal" in name. "One-person" = jobs_reported=eq.1. "DC/MD/VA" = multi-state in.(DC,MD,VA).

# Step 1: Broad query — all staffing firms with 1 employee in DMV
?borrower_state=in.(DC,MD,VA)&jobs_reported=eq.1&naics_code=in.(561311,561312)&order=initial_approval_amount.desc&limit=100

# Step 2: Narrow — only firms with "legal" in the name
?borrower_state=in.(DC,MD,VA)&jobs_reported=eq.1&naics_code=in.(561311,561312)&borrower_name=ilike.*legal*&limit=100

Note: borrower_name is not an indexed filter column — it works here as an additional filter alongside the indexed borrower_state and naics_code filters. At least one fast-lookup column must always be present.

"What are the biggest restaurants in Miami?"

Thinking: "Restaurants" = NAICS sector 7225. "Biggest" = sort by loan amount (proxy for size). "Miami" = city filter requires state co-filter (FL).

?borrower_state=eq.FL&borrower_city=eq.MIAMI&naics_code=like.7225*&order=initial_approval_amount.desc&limit=20&select=borrower_name,borrower_city,initial_approval_amount,naics_code,jobs_reported

"Show me PE firms and hedge funds in New York"

Thinking: PE/VC/hedge funds = NAICS 523999. "New York" = state NY (for the state) or add city filter for NYC specifically.

# All of New York state
?borrower_state=eq.NY&naics_code=eq.523999&order=initial_approval_amount.desc&limit=50&select=borrower_name,borrower_city,initial_approval_amount,jobs_reported,business_type

# Just Manhattan (narrow by city — URL-encode spaces: NEW%20YORK)
?borrower_state=eq.NY&borrower_city=eq.NEW%20YORK&naics_code=eq.523999&order=initial_approval_amount.desc&limit=50

"Find small dental practices in Texas that didn't get their loans forgiven"

Thinking: "Dental" = NAICS 621210. "Small" = low jobs_reported. "Didn't get forgiven" = forgiveness_amount is null. "Texas" = state TX.

?borrower_state=eq.TX&naics_code=eq.621210&forgiveness_amount=is.null&jobs_reported=lte.5&order=initial_approval_amount.desc&limit=50&select=borrower_name,borrower_city,initial_approval_amount,jobs_reported,loan_status

"How many coffee shops got PPP loans in California?"

Thinking: "Coffee shops" = NAICS 722515. "How many" = count query. "California" = state CA.

# Count only (no rows returned)
curl -I "https://newbizbot.ai/api/data/ppp_loans?borrower_state=eq.CA&naics_code=eq.722515&limit=0" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Prefer: count=exact"
# Read the Content-Range response header for the count

"Find IT consulting companies with 10-50 employees in the Tri-state area"

Thinking: "IT consulting" = NAICS 541512. "10-50 employees" = range filter on jobs_reported. "Tri-state" = NY, NJ, CT.

?borrower_state=in.(NY,NJ,CT)&naics_code=eq.541512&jobs_reported=gte.10&jobs_reported=lte.50&order=jobs_reported.desc&limit=100&select=borrower_name,borrower_city,borrower_state,initial_approval_amount,jobs_reported

Note: Repeating the same parameter (jobs_reported=gte.10&jobs_reported=lte.50) is valid PostgREST syntax — both conditions are ANDed together to create a range filter.

"List all nonprofits in healthcare across the Southeast"

Thinking: "Nonprofits" = non_profit=eq.Y. "Healthcare" = NAICS sector 62. "Southeast" = multi-state region. Note: non_profit is not an indexed column, but works as an additional filter alongside the indexed borrower_state filter.

?borrower_state=in.(GA,FL,NC,SC,AL,TN,MS,LA)&naics_code=like.62*&non_profit=eq.Y&order=initial_approval_amount.desc&limit=100&select=borrower_name,borrower_city,borrower_state,naics_code,initial_approval_amount,jobs_reported

"Compare PPP loan totals for restaurants vs construction in Florida"

Thinking: This is an aggregate query — use the RPC endpoint, not the table.

# Restaurants in FL
curl "https://newbizbot.ai/api/data/rpc/ppp_stats_by_industry?p_state=FL&p_naics_prefix=7225" \
  -H "Authorization: Bearer YOUR_API_KEY"

# Construction in FL
curl "https://newbizbot.ai/api/data/rpc/ppp_stats_by_industry?p_state=FL&p_naics_prefix=23" \
  -H "Authorization: Bearer YOUR_API_KEY"

Translation Cheat Sheet

Human concept	API filter
Geographic region	`borrower_state=in.(STATE1,STATE2,...)`
Specific city	`borrower_state=eq.XX&borrower_city=eq.CITY` (state required)
By ZIP code	`borrower_zip=eq.XXXXX`
Industry / sector	`naics_code=like.XX*` or `naics_code=eq.XXXXXX` (see NAICS codes)
Niche within industry	NAICS filter + `borrower_name=ilike.keyword`
Company size	`jobs_reported=eq.N` or `jobs_reported=gte.N&jobs_reported=lte.M`
Loan size	`initial_approval_amount=gt.N`
Forgiveness status	`forgiveness_amount=is.null` (unforgiven) or `forgiveness_amount=gt.0`
First vs second draw	`processing_method=eq.PPP` or `processing_method=eq.PPS`
Nonprofits	`non_profit=eq.Y` (not indexed — combine with an indexed filter)
Business structure	`business_type=eq.Corporation` or `eq.Sole%20Proprietorship` etc. (not indexed — combine with an indexed filter)
"How many"	Use `Prefer: count=exact` header + `limit=0`
"Total/average"	Use RPC: `/rpc/ppp_stats_by_state` or `/rpc/ppp_stats_by_industry`
"Biggest / top N"	`order=initial_approval_amount.desc&limit=N`
"Smallest"	`order=initial_approval_amount.asc&limit=N`

Aggregate Functions (RPCs)

The API provides pre-built RPC endpoints for aggregate queries and paginated lookups. These are optimized and typically return results in milliseconds.

ppp_stats_by_state

Returns aggregate statistics grouped by state.

Endpoint: GET /data/rpc/ppp_stats_by_state

Parameters

Parameter	Type	Required	Description
`p_state`	text	No	Filter to a specific state (2-letter code). Omit for all states.

Examples

# Single state
curl "https://newbizbot.ai/api/data/rpc/ppp_stats_by_state?p_state=CA" \
  -H "Authorization: Bearer YOUR_API_KEY"

[
  {
    "state": "CA",
    "total_loans": 1025643,
    "total_amount": 98765432100.00,
    "avg_amount": 96296.50,
    "total_jobs": 12345678,
    "total_forgiven": 87654321000.00
  }
]

# All states
curl "https://newbizbot.ai/api/data/rpc/ppp_stats_by_state" \
  -H "Authorization: Bearer YOUR_API_KEY"

ppp_stats_by_industry

Returns aggregate statistics grouped by NAICS industry code.

Endpoint: GET /data/rpc/ppp_stats_by_industry

Parameters

Parameter	Type	Required	Description
`p_state`	text	No	Filter to a specific state. Omit for nationwide.
`p_naics_prefix`	text	No	Filter by NAICS code prefix (e.g., `72` for Accommodation and Food Services). See NAICS Code Reference.

Examples

# Restaurant industry in Texas
curl "https://newbizbot.ai/api/data/rpc/ppp_stats_by_industry?p_state=TX&p_naics_prefix=7225" \
  -H "Authorization: Bearer YOUR_API_KEY"

[
  {
    "naics_code": "722511",
    "total_loans": 18432,
    "total_amount": 2156789000.00,
    "avg_amount": 117023.45,
    "total_jobs": 287654
  },
  {
    "naics_code": "722513",
    "total_loans": 15678,
    "total_amount": 987654000.00,
    "avg_amount": 63001.15,
    "total_jobs": 198765
  }
]

# All industries in Texas
curl "https://newbizbot.ai/api/data/rpc/ppp_stats_by_industry?p_state=TX" \
  -H "Authorization: Bearer YOUR_API_KEY"

# Healthcare nationwide
curl "https://newbizbot.ai/api/data/rpc/ppp_stats_by_industry?p_naics_prefix=62" \
  -H "Authorization: Bearer YOUR_API_KEY"

Returns up to 100 industry codes, sorted by total amount descending.

Combining with NAICS Hierarchy

Use the p_naics_prefix parameter at different levels of the NAICS hierarchy to zoom in or out:

# Broad: entire Accommodation & Food Services sector (72)
curl "https://newbizbot.ai/api/data/rpc/ppp_stats_by_industry?p_state=NY&p_naics_prefix=72" \
  -H "Authorization: Bearer YOUR_API_KEY"

# Narrow: just restaurants (7225)
curl "https://newbizbot.ai/api/data/rpc/ppp_stats_by_industry?p_state=NY&p_naics_prefix=7225" \
  -H "Authorization: Bearer YOUR_API_KEY"

# Exact: full-service restaurants only (722511)
curl "https://newbizbot.ai/api/data/rpc/ppp_stats_by_industry?p_state=NY&p_naics_prefix=722511" \
  -H "Authorization: Bearer YOUR_API_KEY"

ppp_loans_by_city

Returns paginated loan records for a specific city, sorted by loan amount descending. Pagination is fast even on high-cardinality cities (e.g., Los Angeles).

Endpoint: GET /data/rpc/ppp_loans_by_city

Parameters

Parameter	Type	Required	Description
`p_state`	text	Yes	2-letter state code
`p_city`	text	Yes	City name (case-insensitive)
`p_limit`	integer	No	Page size (default 25, max 1,000)
`p_offset`	integer	No	Number of rows to skip (default 0)

Examples

# First 25 loans in Los Angeles, CA
curl "https://newbizbot.ai/api/data/rpc/ppp_loans_by_city?p_state=CA&p_city=LOS%20ANGELES" \
  -H "Authorization: Bearer YOUR_API_KEY"

[
  {
    "id": 1234567,
    "loan_number": "3856107205",
    "borrower_name": "ACME CORP",
    "borrower_city": "LOS ANGELES",
    "borrower_state": "CA",
    "initial_approval_amount": 10000000.00,
    "forgiveness_amount": 10000000.00,
    "jobs_reported": 500,
    "naics_code": "722511",
    "business_type": "Corporation",
    "date_approved": "2020-04-10",
    "total_count": 148523
  }
]

# Page 2 (rows 26-50)
curl "https://newbizbot.ai/api/data/rpc/ppp_loans_by_city?p_state=CA&p_city=LOS%20ANGELES&p_limit=25&p_offset=25" \
  -H "Authorization: Bearer YOUR_API_KEY"

The total_count field is included on every row, so you can build pagination UI without a separate count request.

ppp_loans_by_state_sector

Returns paginated loan records filtered by state and one or more 2-digit NAICS sectors, sorted by loan amount descending.

Endpoint: GET /data/rpc/ppp_loans_by_state_sector

Parameters

Parameter	Type	Required	Description
`p_state`	text	Yes	2-letter state code
`p_naics_prefixes`	text[]	Yes	Array of 2-digit NAICS sector codes (e.g., `{"72","62"}`)
`p_limit`	integer	No	Page size (default 25, max 1,000)
`p_offset`	integer	No	Number of rows to skip (default 0)

Examples

# Restaurants (72) in Texas
curl "https://newbizbot.ai/api/data/rpc/ppp_loans_by_state_sector?p_state=TX&p_naics_prefixes=%7B72%7D" \
  -H "Authorization: Bearer YOUR_API_KEY"

# Restaurants + Healthcare (72, 62) in New York
curl "https://newbizbot.ai/api/data/rpc/ppp_loans_by_state_sector?p_state=NY&p_naics_prefixes=%7B72,62%7D&p_limit=50" \
  -H "Authorization: Bearer YOUR_API_KEY"

Note: The p_naics_prefixes parameter is an array. URL-encode the curly braces: {72,62} becomes %7B72,62%7D.

ppp_loans_by_city_sector

Returns paginated loan records filtered by state, city, and one or more 2-digit NAICS sectors, sorted by loan amount descending.

Endpoint: GET /data/rpc/ppp_loans_by_city_sector

Parameters

Parameter	Type	Required	Description
`p_state`	text	Yes	2-letter state code
`p_city`	text	Yes	City name (case-insensitive)
`p_naics_prefixes`	text[]	Yes	Array of 2-digit NAICS sector codes
`p_limit`	integer	No	Page size (default 25, max 1,000)
`p_offset`	integer	No	Number of rows to skip (default 0)

Examples

# Healthcare businesses in Miami, FL
curl "https://newbizbot.ai/api/data/rpc/ppp_loans_by_city_sector?p_state=FL&p_city=MIAMI&p_naics_prefixes=%7B62%7D" \
  -H "Authorization: Bearer YOUR_API_KEY"

# Restaurants + Retail in Chicago, IL
curl "https://newbizbot.ai/api/data/rpc/ppp_loans_by_city_sector?p_state=IL&p_city=CHICAGO&p_naics_prefixes=%7B72,44%7D&p_limit=50" \
  -H "Authorization: Bearer YOUR_API_KEY"

Both ppp_loans_by_state_sector and ppp_loans_by_city_sector include a total_count field on every row for easy pagination.

Counting Rows

For simple row counts, you don't need an RPC function. Use the Prefer: count=exact header with limit=0:

curl -I "https://newbizbot.ai/api/data/ppp_loans?borrower_state=eq.CA&limit=0" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Prefer: count=exact"

The count is in the Content-Range response header: Content-Range: */48523

Known Data Quality Issues

This is raw SBA data loaded as-is. Known issues include:

Issue	Scope	Details
Wrong congressional districts	~226,000 loans	Loans credited to incorrect CDs
Inconsistent city spellings	Widespread	35+ spellings of "Chicago" alone
Invalid state codes	Rare	"FI" instead of "FL", "XX" for unknown
Inflated job counts	~192 loans	Sole proprietorships claiming 500 jobs
Business type mismatches	~23,914 loans	LLC/LLP in name but different classification
Missing demographics	Majority	Most race/ethnicity/gender fields are "Unanswered" or NULL
Rounded loan amounts	~94% of loans under $150K	Amounts end with zero despite being payroll-calculated
NULL forgiveness data	Varies	Some loans still processing or denied

On this page