You are here

Search Engine Customization Prompts Debate on Loan Channeling

Adam Rust's picture

Posted December 20, 2012

Recently it was suggested to me that search engine customization could bring about the unintended effect of sorting the channels through which borrowers shop for loan products.

Even if the name is unfamiliar to you, you probably know of the idea. Most likely, you are aware the Google tends to rank its pages according to search history. If you are someone who has searched for

"general purpose reloadable prepaid card" - an action that would immediately classify you as more of a payments junkie than someone that wrote a request for "Green Dot Card" or even "prepaid card," then it is more likely that you will be directed by google to stories in PayBefore. Indeed, I am a typical example: my search suggestions for "general purpose re.." suggest that I go to GPR cards before "general purpose resume" or "general purpose resin."

Evgeney Morozov has written a review in the New York Times about "The Filter Bubble," a new book by Eli Pariser (the founder of that examines the implications of search engine customization. Pariser's says that the same search request will redirect surfers to different destinations: my input for climate change might go to the IPCC or to the EPA, but another person might end up at Fox News. Moreover, he thinks this should give us reason to worry. What is the prospect for democracy when our modes for understanding are themselves different? It is a point which was given more energy after the contradictory expectations for the popular vote in this year's election. (See Karl Rove and Ohio).

When I heard that at one of those creative economy lunches that loosely masquerades as "work," my thoughts immediately refocused on how this might impact access to financial services. My friend had made her own leap, as well. Her research (no doubt a product of wonky search engine customization) led her to an article that discussed differences in job search strategy. Creative class folks get jobs through contacts, she said, but working class people tend to go to internet job sites. Even within the various internet job sites, there is differentiation in user traffic. For example, readers of show a downward bias relative to LinkedIn or CareerBuilder.

Loan channeling is one ongoing critique within the access to financial services conversation. Observers can take a skeptical stance when a bank provides credit of differing quality through different venues. For instance, one national bank used to offer high-cost installment debt through shops in run-down strip malls and prime mortgages at its branches. While there were some branches in low-income areas - possibly to comply with CRA rules - there were not many of the installment loan stores in upper-income communities.

Search engine customization opens up that possibility in an entirely different venue.

Search results for "loan," by race or ethnicity and by educational attainment.

In order to check that out, I asked a variety of my friends to search for "loan" and "line of credit" on their browsers. So far, the results fail to suggest that customization is leading to any kind of channeling.

I do not see any evidence that customization is leading to any kind of channeling.

The table at the left shows the results for the search term "loan," for four of my contacts. As you can see, three are from protected classes. There is really very little difference. Of note, the top three results in this table and in the next are each from the paid advertisements on Google. The Latino small business owner also used DuckDuckGo. There, he had no advertisements. However, it was not until the eighth entry that he had a lead from a bank. That was for a student loan from SunTrust.

The search engines missed out on student loan opportunities. Three of the four searchers live in a household where one person is currently pursuing a degree.

Searching for "line of credit" resulted in a slightly higher profile of returns. Banks appeared more often. To a degree, the results showed that Google is able to tailor ads by geography. BB&T, for instance, did not appear on "loan" but was ranked at the top of the page rank for each of the three searchers from North Carolina. The Latino small business owner lives in Brooklyn. He had links to US Bank (not in NY), Chase (in New York), and Wells Fargo (in New York). In fact, the most powerful difference could be in the choice of search language itself. Line of credit led to more prime opportunities.

Search results for "line of credit," sorted by race or ethnicity and by educational attainment.