Construction & Engineering Data Provider · Head-to-head
FRED Economic Data (Housing Starts & Construction Series) vs NYC DOB Job Application Filings
Which construction & engineering data provider data fits your job: FRED Economic Data, or NYC DOB Job Application Filings. API, files, or your warehouse. Daily, weekly, or hourly.
FRED Economic Data (Housing Starts & Construction Series)
NYC DOB Job Application Filings
Where the fields line up
No shared field names. These two answer different questions.
| Field | FRED Economic Data | NYC DOB Job Application Filings |
|---|---|---|
series_id | documented | not in this set |
observation date | documented | not in this set |
value | documented | not in this set |
realtime_start | documented | not in this set |
realtime_end | documented | not in this set |
job__ | not in this set | documented |
doc__ | not in this set | documented |
borough | not in this set | documented |
job_type | not in this set | documented |
job_status_descrp | not in this set | documented |
latest_action_date | not in this set | documented |
initial_cost | not in this set | documented |
Coverage, side by side
| FRED Economic Data | NYC DOB Job Application Filings | |
|---|---|---|
| Geographic | United States national primarily; regional and state breakdowns available for many construction series | All five New York City boroughs, down to individual tax lots with census tract, council district and NTA |
| Temporal | January 1959 through the current month for HOUST; varies by series | Latest Action Date January 1, 2000 onward; earliest values in the table begin January 1, 2003 |
What each contains
They tie on 2 attributes. Pick by fit, not by loyalty.
| FRED Economic Data | NYC DOB Job Application Filings | |
|---|---|---|
| Publisher | Federal Reserve Bank of St. Louis | New York City Department of Buildings, published through NYC Open Data |
| Subject | US housing starts, building permits, construction spending and construction employment time series | Every DOB job application filed through Borough Offices, eFiling or the HUB with a Latest Action Date since January 1, 2000 |
| Geographic coverage | United States national primarily; regional and state breakdowns available for many construction series | All five New York City boroughs, down to individual tax lots with census tract, council district and NTA |
| Temporal coverage | January 1959 through the current month for HOUST; varies by series | Latest Action Date January 1, 2000 onward; earliest values in the table begin January 1, 2003 |
| Detail level | One row per observation of a published economic series | One row per job application document, with per-milestone dates and building-level attributes |
| Documented fields | 5 documented columns in the observations view (series_id, date, value, realtime_start, realtime_end) | 95 published columns; 12 carried in the Datadory field dictionary, from job number to gis_latitude |
| Formats | CSV, JSON, XML, XLSX | CSV, JSON, GeoJSON |
| Delivery cadence | Daily, weekly, or hourly - your call | Daily, weekly, or hourly - your call |
| Best for | Housing-cycle modeling, macro and proptech dashboards, spending and payroll trend work | Deal-level prospecting, permit-pipeline analytics, address-level investigation, map layers |
What each does better
FRED Economic Data
Depth in time. HOUST housing starts reach back to January 1959 and run through the current month - nearly seven decades of monthly observations, with seasonally adjusted and non-seasonally adjusted variants side by side. The DOB table only holds jobs with a Latest Action Date of January 1, 2000 or later, and the earliest values in it begin 01/01/2003.
Indicator breadth in one place. Four families sit under one roof: starts, permits, construction spending (manufacturing, residential, nonresidential, private residential) and construction employment and payrolls. The DOB table documents one agency's application pipeline in one city.
Geographic reach. Coverage is United States national primarily, with regional and state breakdowns available for many construction series. DOB filings stop at the five boroughs.
Series shaping done for you. Transformations - change, percent change, annualized percent change, log - apply to any series, frequency aggregates down to annual, and vintage views through ALFRED reconstruct values as they were known on any past date. Revision-aware backtests start shaped, not hand-rolled.
NYC DOB Job Application Filings
Resolution. About 2.72 million rows across 95 columns describe individual job applications: initial_cost and total_est__fee as filed, zoning square footage, stories, height, dwelling units and occupancy before and after, enlargement dimensions, zoning districts. FRED's observations dictionary is five fields wide - series_id, date, value, realtime_start, realtime_end - and cannot describe a single building.
Milestone trails. Each row carries a dated progression - pre-filing, paid, fully paid, assigned, approved, fully permitted, sign-off - so approval cycle times are measurable per job rather than inferred from a monthly index.
Named actors. Records expose the applicant's last name and professional title alongside the owner's name and business name. Prospecting and competitive intelligence live here; FRED serves analysts watching markets, not counterparties.
Geometry included. GIS-derived latitude, longitude, council district, census tract and neighborhood tabulation area come standard, and GeoJSON is a native format. A FRED series carries no point geometry at all.
Scope stated honestly. Filings submitted through DOB NOW are excluded from this table and live in the companion DOB NOW: Build - Job Application Filings dataset - essential to know before counting current activity.
Where they're equivalent
Same industry shelf. Both sit in Construction & Engineering, whose 19 primary records lean heavily on official-sector publishers, and both rank among the slice's six 9-scored datasets.
First-party government records. One aggregates the Census Bureau, BLS, BEA and HUD; the other publishes the Department of Buildings' own case files through NYC Open Data. Neither is an aggregated or resold middleman product.
Tabular, dictionary-documented, timestamped. Each ships stable column dictionaries over tabular files, CSV and JSON included - FRED adds XML and XLSX, DOB adds GeoJSON - and every row answers "when is this true?": an observation date on one side, a latest-action date plus a full milestone trail on the other.
The verdict
Verdict: sample both, pick by fit - they are different instruments pointed at the same industry.
Accept that no row tells you anything about a specific address.
Take NYC DOB Job Application Filings if your question has an address or a name in it. Developers scouting submarkets, contractors prospecting applicants and owners, analysts measuring permit-to-permit cycle times, journalists investigating a filing history since 2000, and any map layer needing real coordinates. Accept city-only geography and no national index.
Short version: if the work starts with "how many units nationally", take FRED. If it starts with "who is building what on this block", take the DOB table. See the data-scientist and developers-builders breakdowns for the industry.
Sample both, pick by fit. See FRED Economic Data · See NYC DOB Job Application Filings
Or take both in one feed
Yes - as the two axes of one construction-cycle workflow rather than as substitutes.
Set the backdrop with the national series: monthly HOUST and PERMIT observations frame whether homebuilding is accelerating or cooling. Then read the city against it: counts and dollar values of DOB job filings by borough and job type show whether New York is leading or lagging the national cycle. Comparing the DOB table's milestone trails against the timing of national turning points turns both into a lead-lag study.
Two cautions from the records. First, there is no shared identifier between them - one keys on series mnemonics and observation dates, the other on job numbers and tax lots - so the bridge is calendar time, joined on your own geography mapping. Second, the DOB table excludes DOB NOW submissions, so pair the companion DOB NOW: Build - Job Application Filings dataset before claiming complete coverage of current activity.
Or take both in one feed: Datadory delivers them alongside the rest of the construction & engineering catalog, normalized to their documented field dictionaries, on the schedule your team sets - daily, weekly, or hourly.
API, files, or your warehouse. Daily, weekly, or hourly.
Fair questions
Is FRED Economic Data (Housing Starts & Construction Series) better than NYC DOB Job Application Filings?
Better for different questions. Both score 9/10.
Do FRED Economic Data (Housing Starts & Construction Series) and NYC DOB Job Application Filings cover the same ground?
Barely. One measures markets; the other itemizes buildings. They complement rather than substitute.
Which dataset goes back further in time?
FRED, by decades. The HOUST housing-starts series extends to January 1959 through the current month. The DOB table holds jobs with a Latest Action Date of January 1, 2000 or later, and the earliest values in it begin January 1, 2003.
Can FRED Economic Data (Housing Starts & Construction Series) tell me about a single building?
No. Its observations dictionary is five fields wide - series_id, date, value, realtime_start, realtime_end - describing published series, not properties. There are no addresses, owners, applicants or coordinates. Building-level questions belong to the NYC DOB filings table.
Which should a mapping or proptech product choose?
The DOB table, decisively. It ships latitude, longitude, borough, council district, census tract and neighborhood tabulation area on every row, with GeoJSON as a native format. FRED carries no geometry at all - though pairing it adds the national housing backdrop to any map-driven analysis.