Extract Inclusionary Zoning Unit counts from PUD applications and Covenants

[NealHumphrey]

@salomoneb this is a ticket as we discussed on the phone for looking into extracting data from PUDs about inclusionary zoning.

Example: This status page at DCOZ for the Atlantic Plumbing buildings (linked from our PUD data source) has a pdf linked under the 'View order' link.

This is the section copied from the 'Benefits' section:

The Applicant proposes to devote an area equal to 15 percent of the density gained through the PUD process to affordable housing for those households whose income does not exceed 80 percent of the Area Median Income as that term is defined by the U.S. Department of Housing and Urban Development. The proposed Project will include a total of 695 apartments on the four parcels. Of these, approximately 14 percent will be studio apartments, 29 percent will be one bedroom units, 40 percent will be one-bedroom with den units, and 16 percent will be two-bedroom units. The Applicant proposes a similar mix of units for the affordable housing requirement, providing 18,800 square feet of affordable housing on the Atlantic Plumbing South parcels and 11,000 square feet of affordable housing on the Atlantic Plumbing North parcel. The Applicant has requested flexibility with regard to the size and type of units. If the allocation of market-rate unit types changes, the allocation of affordable units will change to reflect this allocation. The affordable units will be distributed among floors on Parcels A, B, and C. The units will be affordable for a 20-year term. The Department of Housing and Community Development will determine the price and enforce the affordability of the units through covenants and other legal mechanisms.

Things to note:

• Looks like we will need a manual process to extract expected unit counts from these filing documents.
• Can we set up a script to extract the appropriate paragraph for volunteers to use?
• Should try extracting this data from a few filings to determine appropriate/ relevant fields (i.e. bedroom mix, number of units, floor area, etc.)
• This says DHCD will determine enforcement method. For projects that have already put into place, can we get data from DHCD on affordable units?
• Later in the PDF, it also says the owner is required to record "a covenant in the land records of the District of Columbia, between the owner(s) and the District of Columbia...Such covenant shall bind the Applicant and all successors in title to construct on and use the applicable PUD Parcel in accordance with this Order or amendment thereof by the Zoning Commission." I think these covenants are similarly difficult to obtain in a systematic way, but this may be another way to validate affordable units down the line.
• Does this site by DHCD have all current IZ units in it? http://www.dchousingsearch.org/

Find PUD applications via the zoning map: http://maps.dcoz.dc.gov/zr16 Click the red highlighted areas, and then find the link under 'Zoning commision cases' on the left hand side. This will link to the pages similar to the Atlantic Plumbing case. You can also find a list of all of them that was scraped from this map in our S3 bucket as a CSV file (35 MB)

First steps to complete this ticket:

• [ ] Access 5-10 PUD applications, and find the relevant document that details the outcome of the zoning commission decisions.
• [ ] Note any patterns on how to identify the appropriate document.
• [ ] Extract all paragraphs relevant to the Inclusionary Zoning requirements. Search for terms like 'affordable housing' or 'affordable' and 'Area Median Income'.
• [ ] Make a table of all the relevant data fields for defining the amount of affordable housing in those buildings - e.g. floor area, unit count, bedroom mix, etc.
• [ ] Note any trends in wording that we may be able to use to extract this progmatically.
• [ ] If appropriate, write a Python script that reads the CSV and extracts data from these 5-10 sample cases. It is likely that all we will be able to do with Python is extract the whole paragraph that contains a list of key words, to make it a little easier to do manual extraction later.

[NealHumphrey]

New contributor Emma might be working on this - I followed up with her today to see if so. Will update the ticket with status.

[NealHumphrey]

@emkap01 I'm assigning this to you per our discussion last night! Be sure to add links to the Google Doc notes you're making to this issue so we can keep track of where they are. Thanks!

[emkap01]

This research report explains that we need to scrape specific documents (i.e., "Orders") from all case files related to PUDs that are hosted on the portal, and then do some additional parsing on those docs to figure out which ones might be relevant.

I can design a schema for what we would want to do with the parsed data, but that seems premature given that we don't know whether we have the cycles/bandwidth to scrape the docs we would need.

Theoretically the number of actual PUD cases that could potentially contains orders with useful information in the should not exceed 692, as explained in the report.

[emkap01]

As a next step, it would probably be worthwhile to run a quick check in order to see how many of the properties represented in the (potentially) 692 PUD cases are already mentioned in the Preservation Catalog. Obviously there might still be details in the the PUD documents that would be useful to have even if PUD properties are already in the Catalog, but finding out whether the PUD properties appear to include previously unknown affordable housing inventory would be a good first step.

I looked over the data dictionary docs and it looks like we could match based on SSL (or at least on the first four digits of each SSL value, since that represents the "Square" in which the property is located). I'm happy to take a shot at this but I would need to grab a copy the Postgres db.

[NealHumphrey]

@emkap01 Sounds like a great idea. I just sent you database usage info and login stuff, let me know if you hit any snags connecting!

[NealHumphrey]

@emkap01 - I was looking at DHCD's report page, and found they do an annual report on IZ that includes the quantity of units issued permits / built in each year. But, they don't list out where they are or how they track them. We should see if we can find that out, reaching out to find the right person there.

http://dhcd.dc.gov/page/dhcd-reports most recent available (2015): http://dhcd.dc.gov/sites/default/files/dc/sites/dhcd/publication/attachments/2015%20IZ%20Annual%20Report%20Final%2012-14-16.pdf

They also note that IZ has only been in effect for ~6 years, so we can also use that to exclude some of our PUD data if we still need to go there to get it.

[emkap01]

Good find. One of the footnotes speaks briefly to the sourcing of those numbers:

"The tabulation of IZ units produced annually is based on the Notice of Availability issue date submitted by the developer to DHCD. The production numbers provided in questions 1–5 and 7 do not include financially subsidized affordable housing projects that are exempt from IZ administrative and reporting rules..."

The exemptions would probably involve a fair amount of work to catalog accurately, above and beyond the non-exempt units. But it definitely all sounds like data worth pursuing.

As for the PUD data, I updated the research report to reflect that roughly half of the potential 692 properties represented in pud.csv appear to already be in the Postgres db (based on a crude comparison of the parcel "Square" value between both datasets). Again that doesn't mean the PUD records are not worth more effort to collect and analyze, but if nothing else maybe they should be lower priority than other data, since at a glance it looks like there may not be much additional information about the overall IZ inventory reflected in those PUD cases.

[NealHumphrey]

Update to this - the Affordable Housing Data Set from #194 apparently includes most IZ units (it is only missing those that are part of the ADU program. Just verifying we should put this PDF ruling-based approach on hold until we've fully utilized these other data sources.

[emkap01]

@NealHumphrey I will assume this issue will remain "on hold" unless you think otherwise.