Otis Nash works six days a week at two jobs, as a security guard and a pest control technician, but still struggles to make the $190.69 monthly Geico car insurance payment for his 2012 Honda Civic LX.
“I’m on the edge of homelessness,” said Nash, a 26-year-old Chicagoan who supports his wife and 7-year-old daughter. But “without a car, I can’t get to work, and then I can’t pay my rent.”
Across town, Ryan Hedges has a similar insurance policy with Geico. Both drivers receive a good driver discount from the company.
Yet Hedges, who is a 34-year-old advertising executive, pays only $54.67 a month to insure his 2015 Audi Q5 Quattro sports utility vehicle. Nash pays almost four times as much as Hedges even though his run-down neighborhood, East Garfield Park, with its vacant lots and high crime rate, is actually safer from an auto insurance perspective than Hedges’ fancier Lake View neighborhood near Wrigley Field.
On average, from 2012 through 2014, Illinois insurers paid out 20 percent less for bodily injury and property damage claims in Nash’s predominantly minority zip code than in Hedges’ largely white one, according to data collected by the state’s insurance commission. But Nash pays 51 percent more for that portion of his coverage than Hedges does.
For decades, auto insurers have been observed to charge higher average premiums to drivers living in predominantly minority urban neighborhoods than to drivers with similar safety records living in majority white neighborhoods. Insurers have long defended their pricing by saying that the risk of accidents is greater in those neighborhoods, even for motorists who have never had one.
But a first-of-its-kind analysis by ProPublica and Consumer Reports, which examined auto insurance premiums and payouts in California, Illinois, Texas and Missouri, has found that many of the disparities in auto insurance prices between minority and white neighborhoods are wider than differences in risk can explain. In some cases, insurers such as Allstate, Geico and Liberty Mutual were charging premiums that were on average 30 percent higher in zip codes where most residents are minorities than in whiter neighborhoods with similar accident costs.
Our findings document what consumer advocates have long suspected: Despite laws in almost every state banning discriminatory rate-setting, some minority neighborhoods pay higher auto insurance premiums than do white areas with similar payouts on claims. This disparity may amount to a subtler form of redlining, a term that traditionally refers to denial of services or products to minority areas. And, since minorities tend to lag behind whites in income, they may be hard-pressed to afford the higher payments.
Rachel Goodman, staff attorney in the American Civil Liberties Union’s racial justice program, said ProPublica’s findings were distressingly familiar. “These results fit within a pattern that we see all too often — racial disparities allegedly result from differences in risk, but that justification falls apart when we drill down into the data,” she said.
“We already know that zip code matters far too much in our segregated society,” Goodman said. “It is dispiriting to see that, in addition to limiting economic opportunity, living in the wrong zip code can mean that you pay more for car insurance regardless of whether you and your neighbors are safe drivers.”
The Insurance Information Institute, a trade group representing many insurers, contested ProPublica’s findings. “Insurance companies do not collect any information regarding the race or ethnicity of the people they sell policies to. They do not discriminate on the basis of race,” said James Lynch, chief actuary of the institute.
The impact of the disparity in insurance prices can be devastating, a roadblock to upward mobility or even getting by. Auto insurance coverage is required by law in almost all states. If a driver can’t pay for insurance, she can face fines for driving without insurance, have her license suspended and eventually end up in jail for driving with a suspended license. Higher prices also increase the burden on those least able to bear it, forcing low-income consumers to opt for cheaper fly-by-night providers, or forego other necessities to pay their car insurance bills.
It isn’t completely clear why some major auto insurers persist in treating minority neighborhoods differently. It may in part be a vestige of longstanding practices dating back to an era when American businesses routinely discriminated against non-white customers. It’s also possible that the proprietary algorithms used by insurers may inadvertently favor white over minority neighborhoods.
We have limited our analysis to the four states that release the type of data needed to compare insurance payouts by geography. Still, these states represent the spectrum of government oversight of the insurance industry. California is the most highly regulated insurance market in the U.S.; Illinois, one of the least regulated. In addition, some insurers whose prices appear to vary by neighborhood demographics operate nationally. That raises the prospect that many minority neighborhoods across the country may be paying too much for auto insurance, or white neighborhoods, too little.
This investigation marks the first use of industry payout data to measure racial disparities in car insurance premiums across states. It’s part of ProPublica’s examination of the hidden power of algorithms in our lives — from the equations that determine Amazon’s top sellers to the calculations used to predict an individual’s likelihood of committing future crimes.
Our analysis examined more than 100,000 premiums charged for liability insurance — the combination of bodily injury and property damage that represents the minimum coverage drivers buy in each of the states. To equalize driver-related variables such as age and accident history, we limited our study to one type of customer: a 30-year-old woman with a safe driving record. We then compared those premiums, which were provided by Quadrant Information Services, to the average amounts paid out by insurers for liability claims in each zip code.
In California, Texas and Missouri, our analysis is based on state data that covers insurance claims received, and payouts by, the state’s insurers over the most recent five-year period for which data was available. In Illinois, the data covers a three-year period. We defined minority zip codes as having greater than 66 percent non-white population in California and Texas. In Missouri and Illinois, we defined it as greater than 50 percent, in order to have a sufficiently large sample size.
In all four states, we found insurers with significant gaps between the premiums charged in minority and non-minority neighborhoods with the same average risk. In Illinois, of the 34 companies we analyzed, 33 of them were charging at least 10 percent more, on average, for the same safe driver in minority zip codes than in comparably risky white zip codes. (The exception was USAA’s Garrison Property & Casualty subsidiary, which charged 9 percent more.) Six Illinois insurers, including Allstate, which is the second largest insurer in the state, had average disparities higher than 30 percent.
While in Illinois the disparities remained about the same from the safest to the most dangerous zip codes, in the other three states the disparities were confined to the riskiest neighborhoods. In those instances, prices in whiter neighborhoods stayed about the same as risk increased, while premiums in minority neighborhoods went up.
In Missouri and Texas, at least half of the insurers we studied charged higher premiums for a safe driver in high-risk minority communities than in comparably risky non-minority communities. And even in highly regulated California, we found eight insurers whose prices in risky minority neighborhoods were more than 10 percent above similar risky zip codes where more residents were white.
Judging by how much insurers have had to pay out for accident claims in their Chicago neighborhoods, Nash should be paying less than Hedges, not more.
Over a three-year-period, Illinois insurers have paid out about $172 per car each year in bodily injury and property damage claims in Nash’s zip code, 60612, according to data collected by the state insurance commission. That’s 20 percent less than the $216 per car that insurers paid out for similar claims in Hedges’ zip code, 60657.
But the liability premiums charged by Nash’s insurer, Geico Casualty, in those two neighborhoods actually give a discount to the riskier white neighborhood. In Nash’s neighborhood, Geico charges $409 for annual liability coverage for a 30-year-old woman who is a safe driver, according to insurance quotes provided by Quadrant. In Lake View, Geico charges $338 for the same coverage for the same driver.
For the liability portion of their Geico coverages, Nash is paying $831.34 annually, while Hedges is paying just $549.58, according to their records. Hedges pays less even though he bought higher coverage limits for bodily injury, and his Audi is worth about three times as much as Nash’s Honda. A Geico filing in Illinois indicates that it charges more to insure an expensive car than a cheap one.
Nash said he is accustomed to seeing his neighborhood shortchanged. “When you go to the richer neighborhoods, the red light cameras kind of go away,” he said. “That system is kind of designed for you to fail.”
Geico did not respond to repeated requests for comment.
The disparities persist even in affluent minority neighborhoods. Consider Pernell Cox, a Los Angeles businessman who lives in a wealthy enclave in South Los Angeles sometimes referred to as the “Black Beverly Hills.” His insurer Safeco, a subsidiary of Liberty Mutual, charges 13 percent more for a 30-year-old female safe driver in his neighborhood than in a zip code with comparable risk in Woodland Hills, a predominantly white suburb in north Los Angeles.
“I was surprised by the magnitude” of the price difference, Cox said.
Cox then shopped around and realized he could save nearly $400 a year by switching to Allstate for his two Mercedes-Benzes.
Liberty Mutual, the parent company of Safeco, told ProPublica it is committed to offering drivers “competitively priced car insurance coverage options.”
Individual insurers don’t publicly release their losses on a zip-code level, and have long resisted demands for that level of transparency. As a result, our analysis is based on aggregated losses experienced by almost all insurers in a given zip code in California, Illinois and Missouri, and by 70 percent of insurers in Texas.
The California Department of Insurance criticized this approach. It disputed ProPublica’s analysis and findings on the grounds that an individual insurer’s losses in a given zip code may vary significantly from the industry average. “The study’s flawed methodology results in a flawed conclusion” that some insurers discriminate in setting rates, it said.
To be sure, it’s possible that some insurers have proprietary data that justifies the higher premiums we found in minority neighborhoods. Moreover, in any given zip code, an individual insurer’s losses could differ from the average losses experienced by insurers. But it is unlikely that those differences would result in a consistent pattern of higher prices for minority neighborhoods.
Consider the internal losses that Nationwide disclosed in a 2015 rate filing in California. We compared Nationwide’s premiums charged by Nationwide’s Allied subsidiary to Nationwide’s losses and found that minority zip codes were being charged 21 percent more than similarly risky non-minority zip codes — a greater disparity than the 14 percent we found when comparing Allied premiums to overall state risk data.
The Illinois Department of Insurance also criticized ProPublica’s report. “We believe the methodology used in this report is incomplete and oversimplifies the comparison of rates in minority vs. non-minority neighborhoods,” said department spokesman Michael Batkins.
The Texas Department of Insurance said that it was reviewing ProPublica’s analysis. “It’s important to us that rates are fair to all consumers,” said department spokesman Jerry Hagins. The Missouri Department of Insurance did not respond to repeated inquiries.
Many insurers did not respond to our questions. Those that did generally disputed our results and said that they do not discriminate by race in rate setting. Eric Hardgrove, director of public relations at Nationwide, said it uses “nondiscriminatory rating factors in compliance with each state’s ratemaking laws.” He did not respond to inquiries about our analysis of Nationwide’s internal losses in California.
Roger Wildermuth, spokesman for USAA, said that its premiums reflect neighborhood conditions. “Some areas may have slightly higher rates due to factors such as congestion that lead to more accidents or higher crime rates that lead to higher auto thefts,” he said.
Insurers have long cited neighborhood congestion as a factor in their decision-making. In 1940, a young lawyer named Thurgood Marshall wrote to a friend that he had been denied auto insurance by Travelers. When Marshall complained to the company, he was told that “the refusal was on the basis of the fact that I live in a ‘congested area,’ meaning Harlem, and ‘not’ because I am a Negro.”
Marshall, who later argued and won the landmark school desegregation case Brown v. Board of Education and went on to become a Supreme Court justice, concluded, according to his letter, that, “it is practically impossible to work out a court case because the insurance is usually refused on some technical ground.”
In Marshall’s day, redlining was often defined by refusal to provide loans, insurance or other services in minority neighborhoods. But as those practices became public and controversial — due in part to Marshall’s activism as an attorney for the NAACP — insurers stopped asking applicants to identify their race.
In the 1940s, as part of a bargain to win an exemption from federal antitrust laws, the insurance industry agreed to be regulated by state laws that included prohibitions against discriminatory rate setting. Soon after, following model legislation recommended by the National Association of Insurance Commissioners, most states passed laws stating “rates should not be inadequate, excessive or unfairly discriminatory.” The legislation defines discrimination as “price differentials” that “fail to reflect equitably the differences in expected losses and expenses.”
Of course, the laws didn’t immediately stop discrimination. In a thorough examination of MetLife’s history released in 2002, New York state insurance regulators catalogued all of the ways that the company discriminated against black applicants for life insurance — dating back to the 1880s when it refused to insure them at all, to the first half of the 20th century when it required minorities to submit to additional medical exams and sold them substandard plans.
In the 1960s, as insurers stopped asking applicants to declare their race, MetLife began dividing cities into areas. In minority areas, applicants were subject to more stringent criteria, according to the report. In 2002, MetLife agreed to pay as much as $160 million to compensate minorities who were sold substandard policies.
In the auto insurance industry, similar practices occurred. To this day, most auto insurers base premiums in part on “territorial ratings,” derived from the risk of the area where the car is garaged.
The territorial ratings are “a way of taking into account the conditions under which you are driving,” said David Snyder, a vice president at the Property Casualty Insurers Association of America.
This geographic pricing means that the same driver may be charged different rates depending on the part of town in which he or she lives.
In 1978, Los Angeles County Supervisor Kenneth Hahn pleaded with Congress to rectify the stark inequities of territorial ratings. He said the same good driver would pay over $900 if he lived in Watts, a poor black neighborhood, and just $385 if he lived in predominantly white San Diego County.
“They are being ripped off by the biggest companies in America,” Hahn testified.
But Congress didn’t act.
Bill Corley, who is African American, started his career as a Farmers Insurance agent in West Los Angeles in 1977. He said the discrimination wasn’t obvious on the surface. “Officially, you could write insurance anywhere you wanted to write insurance,” he recalled. But, Corley said, if you had too many clients in low-income areas, Farmers executives “would tell you all the problems that could be associated with that, and you were scared off and intimidated from doing so.”
When he sold insurance in minority neighborhoods, Corley said, the Farmers managers “would nitpick it. They would ask you questions about people’s income levels and questions about neighboring properties — which I don’t really recall ever having to address when I was writing policies in other neighborhoods in the city.” Farmers did not respond to repeated inquiries.
Corley persisted, and eventually established a network of independent minority insurance brokers who worked together to persuade leading insurers to make them agents and sell policies through them. Corley, who now works as an independent insurance agent with offices in San Diego and San Jose, said the increased diversity of agents has improved the business. “Agents and brokers were complicit, and helped to perpetuate redlining, by not making an effort to write policies in those areas,” he said.
Today, some insurers consider other factors beyond the risk of accident payouts in setting rates. Such criteria as credit score and occupation have been shown to result in higher prices for minorities.
Allstate is implementing a new method for tailoring rates to “micro-segments” that appear to be as small as an individual policyholder — a method referred to in the industry as price optimization.
More than a dozen states have set limits on insurers’ use of price optimization, expressing concerns that the technique allows insurers to raise premiums on customers who don’t shop around for better rates. In 2014, for instance, the Maryland Insurance Administration banned price optimization, saying it results in rates that are “unfairly discriminatory.” (In this context, discrimination refers to any pricing that is not related to risk; the effect on minority neighborhoods has not been studied.)
Allstate has disclosed in filings that it is using price optimization in at least 24 states, including Illinois, Missouri and Texas. Allstate spokesman Justin Herndon said the company “uses the likelihood of loss to price insurance which is required by law and specific prices are approved by state regulators.”
In California, when insurers set rates for sparsely populated rural zip codes, which tend to be whiter, they are allowed to consider risk in contiguous zip codes of their own choosing. Often, the companies group these zip codes with similar areas that also have few policy-holders, according to insurers’ rate filings. They then assign lower risk to the entire region than appears to be warranted by the state’s accident data.
However rates are calculated, auto insurance remains unaffordable in many predominantly minority areas of the nation, according to an analysis by ProPublica of U.S. Census data and 30 million auto insurance quotes.
We found that households in minority-majority zip codes spent more than twice as much of their household income on auto insurance (11 percent), compared with households in majority white neighborhoods (5 percent). The U.S. Treasury Department has defined auto insurance as affordable if it costs 2 percent or less of household income.
Consider Kelley Jenkins, a 39-year-old mother of three who lives on Chicago’s South Side. When she was laid off from an office job last summer, she tried to make ends meet by driving for Uber and Lyft. But after two months of sporadic driving, when she was sometimes making only $100 or $200 a week, she couldn’t afford to keep up her $112 monthly auto insurance payments. “I was in a major struggle,” she said.
When she gave up her auto insurance, she lost her driving gigs. Luckily, she soon found a job as a security guard. But she still can barely afford auto insurance, so she bought a bare bones plan from a low-cost insurer.
Jenkins said she would love to get insurance from one of the brand-name companies, but every time she calls for a quote, she realizes, “Oh no, I can’t afford it.”
Over the years, efforts to investigate redlining in car insurance have repeatedly been stymied by the same barrier: the industry’s refusal to make crucial data available.
After the Rodney King riots in Los Angeles in 1992, when people took to the streets to protest the acquittal of policemen who had been filmed beating a black driver, it turned out that about half of an estimated $1 billion in losses from destroyed businesses and homes were not covered by insurance.
California Insurance Commissioner John Garamendi blamed discriminatory practices by the nation’s insurance companies. Touring the battered ruins of the city a month after the riots, he told a New York Times reporter, “I am convinced redlining exists. The bottom line is either you can’t get or can’t afford it.”
Garamendi subsequently approved rules that required insurers to report their market share by zip code. But insurers argued that the data was a trade secret that couldn’t be released to the public. It wasn’t until 2004, after years of legal battles, that insurers lost their case in California Supreme Court.
Also spurred by the Los Angeles riots, several Congressional committees held hearings and began studying the issue of redlining, but were stymied by lack of data. The U.S. General Accounting Office, now known as the U.S. Government Accountability Office, reported in 1994 that an analysis of insurance availability would require insurance companies to begin reporting data at zip code or census tract level nationwide. “Currently available data are insufficient to determine the extent of current problems,” the report stated.
The National Association of Insurance Commissioners also set up a committee to investigate redlining. It didn’t get the necessary data, either.
Robert Klein, who was researching the issue for the association, said in an interview that “the insurance industry opposed the idea of collecting loss and claims data and the NAIC committee sided with the industry and not with me on this point.”
Without data about insurers’ losses, Klein’s report could not determine why premiums were higher in minority neighborhoods — whether the difference was truly because of greater risk there. “Researchers were unable to draw definitive conclusions about the causes of these market conditions,” the report stated.
Insurers say they set prices based on risk but are reluctant to share the data underlying their risk analyses, such as losses per zip code. Publishing data publicly about losses means “you’re creating something that is valuable and you are essentially giving it away,” said Lynch of the Insurance Information Institute.
Texas consumer advocate Birny Birnbaum won a rare victory when, through a public-records request, he obtained data collected by the state insurance commission at a zip-code level.
In 1997, using the information about each insurer’s number of policies, premiums and losses by zip code, Birnbaum published a fiery report naming Nationwide, Safeco, State Farm, USAA and Farm Bureau as among the “worst redliners” in the state because they had much smaller market share in minority neighborhoods than in other neighborhoods.
The insurers sued the Texas Department of Insurance and Birnbaum, contending that the information was a trade secret and making it public had damaged their business. A Travis County district court judge ruled in the insurers’ favor, saying they would “suffer irreparable harm in the absence of a temporary injunction.”
“There were roughly 200 insurance companies in the state. They all sued,” recalled D.J. Powers, who was Birnbaum’s pro bono attorney. “It was the entire auto insurance industry versus me and Birny.”
Since then, Birnbaum has continued to advocate for insurance commissions to collect and publicly release data that can be used for analysis of redlining and other issues. However, to this day, very few states do so. ProPublica filed public-records requests in all 50 states and the District of Columbia seeking zip-code level data about liability claims payouts. Only four states said they collected such data and provided it.
“Regulators are no better equipped to analyze or address these problems than they were 20 or 30 years ago,” Birnbaum said. “If you can’t even monitor the market to identify the problem, you’re certainly not going to be in a position to address the problem.”
On a redlining map of Chicago created by a federal housing agency in 1940, Otis Nash’s neighborhood, East Garfield Park, is colored red for “hazardous.”
“This is a mediocre district threatened with negro encroachment,” the map states. “Most properties are obsolete and the section is very congested.”
The term redlining is sometimes thought to have originated with these maps, which were created for many American cities by the federal Home Owners’ Loan Corporation between 1935 and 1940. The maps were used to assist loan officers in deciding which properties were worth financing.
East Garfield Park was built as a community of townhouses for factory workers. Like much of Chicago’s West and South Sides, it became a predominantly minority neighborhood in the ‘50s and ‘60s as redlining discouraged investment and the city built an expressway and low-income housing projects in the area. Whites fled for the suburbs.
In 1970, East Garfield Park was among the many Chicago neighborhoods swept up in a wave of auto insurance redlining. In an experiment, Illinois had switched in 1969 from traditional auto insurance regulation — in which rates were approved by state regulators before being issued — to a so-called “open rating system” in which companies could issue rates without regulatory permission.
Illinois insurers soon divided Chicago into four territories for rate setting, a scheme that led to higher premiums in black neighborhoods. A group of black insurance brokers banded together to protest what they called a “color tax” that was being levied on black neighborhoods in Chicago.
The issue was severe enough that the U.S. Senate Antitrust and Monopoly subcommittee held a hearing in Chicago to examine it. One witness, undertaker Charles Childs, said the premiums on his two cars for personal use, a Cadillac and a Mercury, had risen from $450 in 1970 to $950 in 1971 and he had to drop coverage for his fleet of undertaker vehicles.
“As rates have been increased in the inner city, they have substantially decreased in essentially white areas,” Millard D. Robbins Jr., the head of the Insurance Brokers Association of Chicago, said at a press conference. “This creates a surtax on blackness and a discount for being Caucasian.”
With black Chicagoans in rebellion, the Illinois legislature declined to renew the experimental open competition law in 1971. But they couldn’t agree on a new law to replace it — so they just allowed the statute regulating auto insurance rate setting and prohibiting discrimination to expire.
Since then, Illinois, home to the corporate headquarters of State Farm and Allstate, has been the only state without legislation explicitly barring excessive or discriminatory rates in car insurance. Illinois does prohibit auto insurers from charging higher premiums to a customer because of his or her physical disability, race, color, religion or national origin.
To address complaints of discrimination from Chicago, lawmakers in 1971 proposed a compromise: They would ban insurers from using the four rating territories in the city.
“The question is: Shall we give the blacks in Chicago a fair break on insurance rates?” said Illinois state Sen. Egbert Groen during the statehouse debate.
In 1972, they passed a law requiring insurers to use a single territorial rate within the city of Chicago for bodily injury coverage. “No one will be helped more than those in the inner city,” predicted Illinois State Rep. Bernard Epton.
But the reality has turned out differently. For many insurers we examined, premiums were high for everyone within the city of Chicago, regardless of risk, compared to the rest of the state. This means that, since Chicago contains one-third of the state’s minority neighborhoods, they are still being overcharged.
And even within Chicago, the law has not prohibited insurers from differentiating prices by neighborhood. That’s because the single-territory rule is limited to bodily injury coverage; rates for property damage can still vary. (Both bodily injury and property damage coverage are mandatory for drivers to purchase in Illinois.)
Consider the premiums that Geico charges to Otis Nash and Ryan Hedges.
In Nash’s zip code, 60612, Geico has set the base rate for property damage insurance at $753 a year, according to the company’s December 2016 rate filing in Illinois.
That’s eight times higher than what Illinois insurers have paid out in property damage claims in that zip code — an average of $91.57 per car in the three years ending in 2014, according to data from the state insurance commission.
By comparison, in Hedges’ neighborhood, 60657, Geico has set the base rate for property damage insurance at $376 a year, according to the same filing.
That’s half of the Geico base rate in Nash’s neighborhood. It’s also only about four times higher than what Illinois insurers have paid out in property damage claims in Hedges’ zip code — an average of $104.45 per car over the same period.
Of course, Geico’s calculations could reflect the unique risk of the insurer’s own clientele that is somehow not reflected in the state averages. But it could also reflect a disparity, unrelated to risk, that punishes a minority neighborhood.
Either way, the $377 disparity between property damage base rates accounts for the majority of the difference in liability premiums paid by Nash and Hedges. The base rate is adjusted by other factors such as age and driving record.
Both Nash and Hedges have about the same amount of property damage coverage for their vehicles.
Despite scraping to make ends meet, Nash bought collision, comprehensive and liability, as well as rental reimbursement, emergency road service and uninsured motorist coverage. “I got everything,” he said, “because you hear so many horror stories.”
He’s dependent on his car. He needs it to go to work, to shop for groceries now that the local pharmacy closed in his neighborhood and the dollar store burned down, and to take his 7-year-old daughter out to the suburbs where she can ride her bicycle in a park that is safe from crime.
“I don’t even walk up and down the block with my daughter,” Nash said, adding that it’s not unusual in the summer to “hear gunshots during the day.”
Nash said he is working with a financial adviser to cut back his expenses so he can make his rent payments. Still, he’s reluctant to give up any of his car insurance. “I would choose the rent over my car, but that would be playing with fire,” he said.
Hedges has even more coverage than Nash, but it is not a financial burden for him.
Hedges’ premium recently went up after his husband got into an accident. But even after the incident, their combined price of $115.37 a month for two cars is lower than the $190.69 a month that Nash pays for just one car. When told about the difference in prices, Hedges said it seemed unfair.
“It’s an unfortunate reflection of where we are in the corporate world and how we treat each other,” Hedges said.
Nash agreed: “Why would you go to the most poor communities and charge more?”
This story is not subject to our Creative Commons license.