Leveraging Open Data platforms to improve government policymaking

This post provides the methodology behind the following post: Rebuilding our small businesses.

I pulled business data from Data SF, the City’s Open Data portal, which publishes a list of every business registered in San Francisco. The 2,349 businesses represent the upper bound of an estimate. Nonetheless, the process of using Open Data portal to inform policy making, demonstrates how policymakers should be using data to drive policy making. Data-driven policymaking enables us to identify small businesses most at economic risk from a shut down in hospitality, traffic, and retail. It also helps us ask the right questions to move forward.

Small business data in San Francisco

I was able to start this process without access to all of the City’s data and resources. It wasn’t that difficult. I pulled one data set from Data SF, but evaluated it in two ways to get an idea of how many vulnerable businesses were in District 3. Specifically, I used the Registered Business Locations San Francisco (RBLSF) dataset.

For my first approach, I filtered the RBLSF dataset to identify those businesses that were still open, that operated in District 3, and then culled that list by the type of business using the NAICS number. Of the 18 NAICS numbers provided by the City, I selected what I *think* will be the most impacted industries: Accomodations, Administrative and Support Services, Food Services, and the Retail Trade. Combined, these industries made up 2,205 companies out of 11,429 total. Or 19.2%.

For the second approach, I filtered the data online (using DataSF’s filters) before downloading the data. I filtered for businesses in District 3, then by the Neighborhood Analysis Boundaries, and finally I culled the data again using the NAICS codes as above. The results were pretty similar: 2,349 companies out of 10,878 total or 21.5%. A slight decrease in the total number of businesses and an increase in the total number of businesses in the NAICS codes used, but within reason.

To be sure, this is just a quick cut at the numbers. I only included the four NAICS listed above as a proxy for size/relevance. I excluded, for example, companies that were defined as “Information” companies. That said, the error could go either way. This is a good example of what type of data that would be useful.