The key objective of the Consumer Reports/ProPublica investigation was to test the auto insurance industry's explanation of insurance pricing—namely, that the higher average rates charged to drivers in minority neighborhoods have nothing to do with racial or ethnic factors, but simply reflect higher risk levels associated with individual drivers in those neighborhoods.
We first needed to compare the amount of insurance claim payouts in minority neighborhoods to payouts in other neighborhoods. That information was not publically available at the time, so ProPublica filed a Freedom of Information Act request with regulators in all 50 states seeking zip-code-level data on liability claim payouts.
We then subjected the data to a first-of-its-kind statistical modeling to compare prices paid by drivers in minority neighborhoods to those paid by drivers in equally risky white neighborhoods.
The finding: even when risk levels were the same, customers in predominantly minority neighborhoods were charged more by some insurers.
Insurers closely guard information about their claims payouts by zip code. We had to find data through public and private sources. To compare this payout data to company premiums, CR had to buy data from Quadrant, a for-profit that collects information from almost all insurance companies. To supplement the premiums data, CR also had to license a proprietary database of insurance regulatory filings from another company.
We also had to clean the data, which involved parsing PDFs that were tens of thousands of pages long, and had to write software to analyze each insurers' filing. Finally, we had to determine how to statistically compare risks and premiums. ProPublica's data journalists and CR's statisticians spent months debating and developing different approaches to the analysis before deciding on the final methodology. (The methodology and modeling was also vetted with experts unaffiliated with CR or ProPublica.)
In September 2017, following an investigation prompted by the article, California regulators announced they were requiring Nationwide and USAA to adjust their auto insurance rates. Nationwide and USAA are two of the 10 largest auto insurance providers in the country by market share.
The department of insurance said that the adjustments would largely erase the racial disparities that ProPublica and CR found in the two companies' pricing.
California was not the only state to respond. In Illinois, one of the least-regulated markets, two state senators, Jacqueline Collins, D-Chicago, and Daniel Biss, D-Evanston, proposed barring car insurers from using a person's zip code to determine premiums. The investigation found 33 of 34 companies analyzed in Illinois were charging at least 10 percent more, on average, for the same driver in minority zip codes than in comparably risky white zip codes. Two weeks after the report came out, six Democratic members of Congress wrote a letter citing the study and urging the Treasury Department to appoint a director for the Federal Insurance Office (FIO), which monitors insurance pricing and availability in minority neighborhoods.