Several days ago, the U.S. Small Business Administration (SBA) announced the release of detailed loan-level data regarding the loans made under the Paycheck Protection Program (PPP). The disclosure covered 4.9 million PPP loans made to business and nonprofits.
The data included business names, addresses, NAICS (industry) codes, zip codes, business type, demographic data, non-profit information, lender names, jobs supported and loan amount and accounted for nearly 75% of loan dollars approved.
PPP loans are aimed at keeping employees on the payroll. According to the SBA, the program has supported 51 million jobs. The low-interest loans can be forgiven if businesses meet certain conditions, such as retaining or rehiring employees and maintaining salary levels.
As federal and local governments collect more data and use it to make important decisions, we’ve seen the proliferation of publicly accessible datasets. However, in most cases, the raw data itself is not comprehendible or actionable for the vast majority of citizens. We are forced to rely on journalists or other ‘data storytellers’ to translate this data into something relevant to us. We may also find our leaders acting as those storytellers to spotlight the insights in the data. Both of these approaches come with risk.
Personally, I believe a key part of our mission at Narrative Science is to promote the use of our products to help every person make sense of the world around them through data. Selfishly, I wanted to take the time to understand this data. I was curious about what industries received the most loans. I wanted to know how top-heavy the distribution of money was in terms of populated states. I was interested in which banks were making the most loans.
I don’t consider myself a ‘data analyst’ by trade, but I know my way around a CSV of loan data. The moment I opened up the file I realized a perfect chance to showcase how powerful Quill can be for this purpose. However, the first thing I did was my own little data enrichment. The spread of loans across industries (i.e., restaurants, lawyers, hotels, etc.) was something I was curious about, but the data only had NAICS codes. I went to the Census Bureau website and grabbed a file that mapped those ‘codes’ to ‘industries’ and joined in the table.
Even after that, I wasn’t going to get what I needed from snooping around a giant spreadsheet. So I decided to organize some of my thoughts into a Microsoft Power BI dashboard. I created some bar/column charts to break down the # of loans made by industry, business type, state and city, lender and loan amount. I took advantage of using filters to play around and see charts change.
To be honest, even after this exercise I was still overwhelmed with the amount of data I had access to and the amount of exploration I could do. Am I technically capable of doing it? Sure. Did I really want to? No. This is where Quill came in. I took 2 minutes placing Quill narratives into my Power BI report and immediately I had some bullet points telling me EXACTLY what I wanted to know.
- By Industry: The top 12 industries account for a quarter of overall # of businesses.
- By Lender: JPMorgan Chase Bank, National Association (36,699) is almost 240 times bigger than the average across the 4,318 lenders.
- By State: # of businesses is relatively concentrated with 75% of the total represented by 19 of the 56 states (34%).
And from there, I could click on items in the story (“The # of businesses was driven by Full-Service Restaurants with 33,608 …) that were interesting to me to get a deeper look. This would give me new narratives to read and also different views of my report to explore.
Over the last decade or so there has been a lot more data provided by governments to citizens, and this is a great thing. However, a table of civically-relevant data is not an end unto itself. Most citizens do not have the tools needed to find the insightful and actionable bits on their own. When we rely on the government or media for that part, we become susceptible to any bias or agenda that may come along.
At Narrative Science, we strive to empower everyone to comprehend datasets like this and make judgements and decisions for themselves. In this instance, Quill for PowerBI was the right tool for the job.