The Historical Commission Now Denies 40% of the Applications That Reach a Hearing, Up From 1 in 7 Before 2019

If you own a Philadelphia rowhouse in a historically protected area, and you want to put in a window that does not match the Historic Preservation standards, you'll need to apply for an exception. The good news is that if your application is reviewed by the staff and has no errors, it will be approved 97% of the time.

If your application is contested enough to reach a hearing, it goes in front of thirteen commissioners, where it will be denied two out of five times (40%), a massive increase from the 14.4% average the Commission ran at for the 35 years before 2019.

The 90% Is Real and It Is Not the Claim It Sounds Like

The Commission's homepage says it "approves over 90% of building permit applications without referral to a committee." That sounds awesome, i.e., a city bureaucracy that won't prevent you from fixing up your own home. And it's kind of true.

Roughly 97% of applications that trigger preservation review are handled by staff planners with no hearing. For non-routine things (i.e., the stuff you hear about in the news), they're reviewed by the Commission. That set of requests goes to a body that has become MUCH less permissive in recent years.

The Denial Rate Went From 14.4% to 40.8% in 2019 and Never Came Back

My data goes back to 1956 and records what the Commission decided, but only some of that data records the actual vote counts of the Commissioners. Before 1980, there are near zero actual vote counts. It's really only around 1984 that we get better data counting how many votes each application got. Before that, the outcome is something you have to read out of the narrative, which is a guess wearing a lab coat.

Starting in 1984, I grouped the applications by year and counted the share that came back denied. The lower panel of the chart is the recording share: the share of decisions where the minutes record a vote at all. That's the sample my count comes from.

From 1984 through 2018, the average denial rate was 14.4%, each year counted once rather than weighted by how many votes it had. The worst single year in that stretch was 29.9%.

Then it moved. Every year since 2019 has been higher than every year before it, and the average is 40.8%. The year right before the break, 2018, was 7%. That's not drift. Something changed in 2019 and it stayed changed.

The Minutes Were Public the Whole Time, and the 90% Points the Other Way

None of this was hidden, which is the strange part. The minutes have been on the city's website the whole time. There are roughly 1,700 of them, most of them scanned images from years when nobody wrote down a vote. No reporter is going to read a few of them and come away with "the Commission got stricter." The one number the city does publish, the 90%, points the other way.

What I thought was that somebody had finally made the minutes readable, and that the first thing anyone does with a readable record is find out what it actually does.

In 2019 the Denials Quadrupled and the Rules Were About to Loosen

The break lands in 2019. The denial rate went from 7% the year before to 30.7%. The law moved the other way that same year: Mayor Kenney signed three bills in November 2019 that loosened the preservation rules, waiving parking requirements triggered by a new use, allowing accessory dwelling units inside existing houses and garages, and letting qualifying adaptive-reuse projects proceed by right without a variance (Bills 190611, 190612 and 190613). They took effect the following January, so the rules were looser after the break than before it. That rules out the simplest explanation for the denials. The minutes are tallied by year, so they can't tell you which month the denial rate moved. I'd put the timing at a coincidence and nothing more.

One candidate I can put numbers on is the recording share itself. It jumped from 45.3% in 2018 to 81.5% in 2019, the same year the denial rate jumped. If the cases that newly got written down were different from the ones that were already, some of the increase could be an artifact of the record instead of the Commission's behavior. I can't settle that from the minutes. But the recording share has stayed between 79% and 88.5% every year since, so the post-2019 average is measured consistently, even if 2019 isn't comparable to what came before.

Beyond that, I haven't found an explanation, and I don't think one is in the public record. It could be membership: thirteen commissioners, and a change in who they are is a change in what gets built. It could be volume: as owners worked out which applications were winnable, the mix that reached the Commission shifted toward harder cases. It could be how the post-2019 applications are drafted, which I can't check from the minutes. Any of those is a guess wearing a lab coat again, which is a phrase I have now earned the right to use twice.

The system isn't broken. It's just not the system its own website describes. If you own a building that got denied, you're in the 40%, not the 90%, and the number moved in 2019 for reasons the minutes don't explain.

Notes, Sources, and Methodology

Decision records come from the minutes_decisions file in the data-current release of Brendan Illis's phc-document-repository. That repository is private, so you can't pull the file and audit my counts directly. The filter is the part that matters if you want to check the number against the underlying minutes. Rows are filtered to decision_type = 'alteration_review', then to rows carrying a recorded vote_tally, and grouped by year. Years with fewer than 10 recorded votes are dropped. The yearly denial rates are then averaged unweighted, counting each year once rather than weighting it by how many votes it had.

The 90% figure is from the Commission's own website. The application process described in the first paragraph is from the city's Get approval for work to a historic property page. The ~97% staff-review figure is the Philadelphia Inquirer, July 30 2025's own characterization of how the city handles review, not a statistic the city publishes.

Three caveats. The decisions that carry a recorded vote are not the same set every year. Recording was jumpy before 2019 and steady after it, so the pre-2019 average rests on a narrower and differently-composed sample. Read it as indicative rather than exact. Outcome labels are machine-extracted and imperfect. Illis's own benchmark against hand-labeled samples puts outcome accuracy near 67%, and I haven't independently re-labeled a sample of the modern denials. And applications resolved through litigation or handled directly by staff often never reach a public meeting at all. So the figures here are the share of applications the Commission decided by recorded vote, not the share of all applications. That is a much less flattering denominator to compute. (The recording share in the lower panel of the chart is a percentage of the cases that reached a decision, so it is not the complement of the 97% staff figure.)

https://nickjain.com/blog/phc-denies-two-in-five