Build With Tango: Which agency accounts are actually worth your attention

This is the first in a short series on the new budget endpoint. The opening question for any agency you might chase: of all the money it spends, how much of it is even contractable?

The problem

An agency's topline budget is a terrible proxy for opportunity. The Department of Health and Human Services obligated over two trillion dollars in FY2024, and a contractor should basically ignore almost all of it. Most of it is Medicare and Medicaid spending: transfers and benefit payments that never become a contract. Buried in there are a handful of accounts that do buy things: labs, IT, medical countermeasures, program support. The trick is telling them apart without reading hundreds of budget justifications.

The query

To help make this possible, Tango has precomputed a metric we call contract_share_of_obligated_capped (we know! clever, right?), which represents the portion of obligated dollars that are related to contracts, bounded so deobligations and refunds can't push the ratio outside [0, 1]. We also allow users to rank an agency's accounts by the metrics.

So, for example, HHS is agency code 075. Let's explore it using the tango-python SDK:

from tango import TangoClient

client = TangoClient()  # reads TANGO_API_KEY from the environment

rows = []
page = 1
while True:
    resp = client.list_budget_accounts(
        fiscal_year=2024,
        agency_code="075",  # HHS
        page=page,
        limit=100,
        shape=(
            "account_title,obligated_total,contract_obligated,"
            "contract_share_of_obligated_capped"
        ),
    )
    rows += resp.results
    if not resp.next:
        break
    page += 1

# Keep accounts that actually move money, rank by contract share
material = [
    r for r in rows
    if r.get("obligated_total") and float(r["obligated_total"]) > 2e8
    and r.get("contract_share_of_obligated_capped") is not None
]
material.sort(key=lambda r: float(r["contract_share_of_obligated_capped"]), reverse=True)

What comes back

The accounts that actually buy things are, counterintuitively, the small ones. The giant budgets are often pass-throughs:

HHS account (FY2024)	Obligated	Contract share
Program Management	$7.43B	70%
Public Health and Social Services Emergency Fund	$7.43B	54%
Services and Supply Fund	$2.76B	58%
Administration for Strategic Preparedness and Response	$2.57B	84%
Health Care Fraud and Abuse Control	$2.57B	55%
Quality Improvement Organizations	$1.18B	91%
…
Grants to States for Medicaid	$690B	1%
Federal Supplementary Medical Insurance Trust Fund	$526B	~0%

The two largest accounts in the agency, which combine for over a trillion dollars, involve essentially 0% contract spending. Meanwhile the real buying happens in a cluster of $1–7B accounts you could miss entirely if you sorted by size. One field, one sort, and the list of accounts worth a capture team's time drops from "all of HHS" to about eight.

What you'd build with it

So what? Knowing which agencies actually have budget for contracts is useful for a variety of workflows, including:

• An agency triage step. Before you build a pipeline for a new agency, rank its accounts by contract share and throw out everything under ~10%. You've just scoped your research to the accounts that can actually award work.
• A portfolio sanity check. Map the accounts you already chase against their contract share. If you're tracking a 1% account, you're tracking a transfer program.
• A cross-agency screen. Drop the agency_code filter and rank the whole government, then filter by bea_category or subfunction_code to find contract-heavy accounts in the mission area you serve.

Field definitions are in the Budget API reference. If you've never run this against your own coverage agencies, it's worth twenty minutes on the free tier.