New search for zip-code level rental data

[santiagohermo]

In this issue we want to search for new sources of rental data. We want to get a list of potential sources, including (if possible) links to website, pricing, and time and geographical coverage.

[santiagohermo]

Who can tackle this issue @gabrieleborg @diegogentilepassaro? The goal of the task is to get a sense of some options only, so only the table with alternatives is required (which shouldn't take too long).

Some options to add to the table:

• HouseCanary
• RealtyMole. See this blog post

[gabrieleborg]

Thanks @santiagohermo! I read your text and will look into this if it's ok with @diegogentilepassaro

[santiagohermo]

Following our off-GitHub discussion I assigned you here @gabrieleborg.

Could you also take a look at the county-level FMR data? It would be good to know what they represent and how they are computed, as we might want to use it (I opened #255 to download it).

[gabrieleborg]

Hey @santiagohermo @diegogentilepassaro,

I have looked a bit into the two potential new data sources for zip code data.

HouseCanary (HC)

• HC has a very rich set of housing market data: home valuations, rental data, forecasting info, and market analysis;
• For rental data, they specialize in single-family home rentals. They provide individual property data and aggregated data at various levels - down to block level - including zip code.
• This is a data-broker company whose customers are mostly professionals and real-estate companies, so they look very protective of their data;
• The only potential way I see to get this data is to contact them as researchers and try to ask whether they’d give us the data. For this reason, I think Santi should contact them, with his brown account. If they don't agree, we could inquire how much would cost to get data at the zip code level, and how rich is their historical data.
• They have a very rich API where you can request any type of data;
• Specifically, I think the zip/details endpoint is what we would ideally want:
• Additionally, they also have data at the block, block-group, metro, msa, and state level;
• If we decide to contact them, I drafted a possible email we could send (with chatGPT help). Feel free to edit/modify:

  Subject: Request for Zip-Code-Level Rental Data for Research Purposes

  Dear HouseCanary Team,

  We hope this email finds you well. We are a team of economists conducting research on the impact of minimum wage legislation on rental prices across the United States. As we strive to enhance the quality of our research, we came across HouseCanary and were impressed by the comprehensive and detailed rental data that your company provides.

  We are writing to inquire over the possibility to access zip-code-level rental data for our research. We believe that HouseCanary's data would greatly contribute to improve our study and to help us more insightful conclusions on the effects of minimum wage legislation on rental prices. This has important implications for labor markets, housing affordability, and social welfare, and we believe could potentially benefit your company’s products too.

  We understand that HouseCanary's business model is based on selling access to data, but we humbly request your support for our research, purely for academic purposes. We assure you that the data provided will be used exclusively for research purposes and will be treated with the utmost confidentiality. We are also committed to appropriately citing and acknowledging HouseCanary in our research papers and publications as a source of the data, which we believe will generate positive publicity for your company.

  We are passionate about our research and believe that HouseCanary's data would significantly enhance the rigor and validity of our findings. We are excited about the potential collaboration and the opportunity to collaborate with your company.

  Thank you for considering our request. We look forward to your positive response. Please do not hesitate to contact us if you require any further information or have any questions.

RealtyMole (RM)

• For this data source, I mostly followed the blog post @santiagohermo linked (thanks that was great!);
• RM also has a very nice data API. To access it, I subscribed to RapidAPI, a platform that allows you to access many APIs;
• RM has data on zipcode-level median, min, and max rent by no. bedrooms (0-5), with also the number of listings;
• PROBLEM: no sqft. info, so we cannot compute rent/sqft;
• The monthly time series goes back to 2020-04 only;
• I adapted the script from the blog post and created an initial python script that - when we decide what to do (SEE NEXT POINT) we can run to get all the data;
• Of course, we can't download all zipcode-level data for free. RapidAPI provides different pricing options, so I made some quick calculation on how much it would cost us to get data for all zipcode (I think 1 zipcode = 1 API request). Since it's a monthly subscription, I computed how much it would cost us to download the data in 1, and 2 months:

[santiagohermo]

Thanks @gabrieleborg! That's really useful. My reactions:

• HouseCanary: Did you see anything on what time-frame is available? I'd be happy to meet with them, but honestly it seems that it's going to be expensive and timely to get.
• RealtyMole: Looks feasible! And even with its limitations I see it as a useful robustness check that would address several concerns.
  • Note that we don't really need all ZIP codes, only the urban ones (should be around 11K according to our stats table. We can cut even more by dropping ZIP codes in states with no MWs). Thus, it might be enough with one month of the Mega subscription. I could ask for some money at the Bravo Center, but if it's for data it's a pain as it has to be approved by someone beyond the department. Could we pay for it ourselves?
  • It's not great that they don't have square feet. Is there any way we could get other estimates beyond minimum and maximum, like the 25th percentile? Anyways, it seems that we could compute something like the average rent per bedroom to proxy for square feet.
  • Sadly, the timing does not line up very well either. We could probably use changes in the MW in 2022-23 only (to avoid covid concerns). We may need to extend our data a little bit.
  • Really useful to know the number of rentals and the composition (how many with 1 bedroom, 2 bedrooms, etc)
  • Using your code I checked the data for zip code 02906, here in College Hill. The output looks really cool! Example:
  • Finally, if we get it it would be useful to get to know a little bit more on how it's computed. But since it'd be a robustness check I don't think we need to go into that much detail.

fyi @diegogentilepassaro

[santiagohermo]

Some quick plots for 2-bedrooms in the zip code 02906

averageRent	totalRentals

You left right on time, rents are getting high!!

[santiagohermo]

For this issue we decided the following:

• If the AEJ Applied accepts the paper, we will probably try to get this data.
• If not, and we aim for JUE, we will not get this data.

Leaving issue open for now.

[santiagohermo]

Looks like we don't need this for now. Could someone close this issue, delete the branch, and make a brief summary comment in case we need to take this back in the future? @diegogentilepassaro @gabrieleborg

[gabrieleborg]

Closing this issue. The goal was to find alternative zip code-level rental data to support our analysis. Potentially, we would ideally want data with number of rentals as well.

[santiagohermo]

Data at the city level since 2017: https://www.apartmentlist.com/research/category/data-rent-estimates