What Normal Looks Like in Philadelphia Permit Data

Philadelphia averages 3,938 permit filings a month with a standard deviation of 354, so a month has to move about 708 permits, near 18%, before it means anything at all. That figure does not exist anywhere in the public coverage of the Philadelphia construction market, which is why almost every claim about the market rising or falling is made without a way to check it. Below is the baseline, drawn from 19 complete months of our own Philadelphia file, along with the separate threshold that applies to each trade.

TL;DR

• Philadelphia runs 3,938 permits a month, standard deviation 354, across 19 months.
• A citywide move under roughly 18% is ordinary variation, not a market signal.
• Every trade has its own threshold: 17% for building, 39% for demolition.
• Filings are down 4.2% year over year, which is smaller than the monthly noise.

What is a normal month in Philadelphia?

Between 3,300 and 4,600 permits, with the middle of the distribution near 3,900.

Here is the full series from our Philadelphia file, January 2025 through July 2026. August 2026 is excluded because it is incomplete at the time of writing. The firm column counts distinct contractor names appearing on that month's filings, which is a useful companion number and one almost nobody publishes.

Month	Permits	Filing firms	Permits per firm
2025-01	3,998	1,061	3.8
2025-02	3,592	1,009	3.6
2025-03	4,487	1,195	3.8
2025-04	4,299	1,095	3.9
2025-05	4,120	1,176	3.5
2025-06	4,012	1,114	3.6
2025-07	3,961	1,030	3.8
2025-08	3,811	1,053	3.6
2025-09	3,769	1,060	3.6
2025-10	4,585	1,148	4.0
2025-11	3,323	982	3.4
2025-12	3,599	1,023	3.5
2026-01	3,391	963	3.5
2026-02	3,484	1,025	3.4
2026-03	4,337	1,069	4.1
2026-04	3,988	1,099	3.6
2026-05	4,033	1,077	3.7
2026-06	4,112	1,152	3.6
2026-07	3,924	1,111	3.5

Mean 3,938. Standard deviation 354. Coefficient of variation 9.0%.

The shape is seasonal and fairly consistent between the two years. March through June are the strong months. November through February are the weak ones. Anyone comparing a February to a May and calling the difference growth has measured the calendar.

How much can filings move without meaning anything?

About 18% citywide, and considerably more than that inside a single trade.

Two standard deviations is the conventional working line, and it is the one worth writing on a whiteboard before you look at a number. For the citywide series that is 708 permits. Per trade, the arithmetic changes sharply, because smaller series are noisier:

Permit type	Monthly mean	Std. dev.	Move needed to matter
Building	1,472	128	±17%
Plumbing	1,139	153	±27%
Electrical	715	79	±22%
Mechanical	414	50	±24%
Fire alarm	162	27	±33%
Demolition	36	7	±39%
All types	3,938	354	±18%

A demolition contractor who watches monthly counts will see a 30% swing regularly and it will mean nothing. A building-permit-driven business seeing the same 30% swing is looking at something real. Same city, same file, same month, opposite conclusions.

Why is the citywide number the worst signal for your trade?

Because aggregation cancels noise, so the citywide series is the steadiest thing in the file and therefore the least representative of what any one contractor lives through.

This is the part that gets the reasoning backwards in most market commentary. The citywide coefficient of variation is 9.0%. Every individual trade except building is above it: plumbing 13.4%, electrical 11.1%, mechanical 12.1%, fire alarm 16.7%, demolition 20.5%. When plumbing has a soft month and mechanical has a strong one, the citywide total absorbs both and reports calm.

So the headline number is smooth by construction. Quoting it to a plumbing contractor, whose real month-to-month experience varies half again as much, describes a market nobody actually works in. The same logic applies to the trade split we published in our breakdown of Philadelphia trade permits and contractor coverage: the trades behave like separate markets that happen to share a filing system.

What happened between October and November 2025?

The largest swing in the file, and it had no regulatory cause.

October 2025 was the busiest month at 4,585 and November 2025 the quietest at 3,323, a fall of 27.5% between adjacent months. There was no fee change, no code deadline and no Department of Licenses and Inspections rule taking effect in that window. The 2021 code transition ran on a different clock entirely, and produced no filing rush of its own.

Decomposed, the drop is not one event:

Permit type	Oct 2025	Nov 2025	Change
Building	1,612	1,253	-22.3%
Plumbing	1,451	941	-35.1%
Electrical	878	571	-35.0%
Mechanical	436	388	-11.0%
Fire alarm	170	129	-24.1%
Demolition	38	41	+7.9%

Plumbing and electrical carried the fall. Mechanical barely moved. Demolition rose. A single citywide figure of -27.5% is wrong for every trade in that table, in one direction or the other.

The firm count is the more interesting half. Distinct filing firms fell from 1,148 to 982, down 14.5%, against a 27.5% fall in permits. Roughly 850 firms filed in both months. This was the same population of businesses filing fewer permits each, not a market losing participants, and those are different problems with different responses.

Is Philadelphia construction actually slowing down?

Mildly, and by less than the file's own noise.

January through July 2026 produced 27,269 Philadelphia permits against 28,469 in the same seven months of 2025. That is a decline of 4.2%. It is real, it is consistent across most months, and it is also smaller than the 9.0% coefficient of variation of the monthly series.

The honest description is a flat market with a slight downward tilt. January is the one month that moved meaningfully, 3,998 down to 3,391, a fall of 15.2%. Every other month-on-month-a-year-ago comparison sits between -7.2% and +2.5%.

If your pipeline is down 4% this year, Philadelphia is the reason and it is not you. If it is down 25%, the market is not the explanation and something closer to home is.

What can this baseline not tell you?

Four things, and the fourth is the one that trips people up.

It does not measure review speed. The Philadelphia feed reports issued permits, and decision_date equals app_date on every record, so no elapsed processing time exists to measure. It does not measure job value: project_cost is empty across all 78,592 of our Pennsylvania records, so none of this can be expressed in dollars. It does not describe Pennsylvania, since we hold 3 of the state's 67 counties and 99.4% of the file is Philadelphia. Reading and Allentown hold a few hundred records between them and cannot support a monthly series at all.

The fourth is subtler. A permit count measures filings, not construction. A single large project can file a building permit and four trade permits, and lands in this data as five records. That is why the firm count matters as a check: it moves differently from volume, and when the two diverge, the divergence is the story.

Underneath all of it sits the thing that makes any of this worth tracking. A permit is a signal about the property owner, not the contractor. The baseline just tells you when the number of those signals is doing something unusual, and 94.9% of our Philadelphia records carry an owner name to act on.

How permits.llc fits in

We hold 78,592 Pennsylvania permit records, 99.4% of them Philadelphia, with 86.1% carrying a contractor name and 94.9% an owner name, current to within about a day. All six active permit types are present and separable, which is what makes a per-trade baseline possible here and not in most of our other files.

You can download the Philadelphia permit data and rebuild every table above against your own trade filter, or start from our Pennsylvania county coverage to see exactly where the file is dense and where it is not. If you would rather be told when something crosses your threshold than check it yourself, permit alerts run the monitoring for you.

One suggestion, freely given. Write your trade's threshold down before you look at next month's number. It is a great deal easier to be honest about a figure you have not seen yet.