This article outlines a clash between researchers over two separate studies which illustrated opposite trends in appraisals. The first study, conducted by CoreLogic in June 2019, concluded that there existed significant evidence of implicit bias in the form of racial and ethnic disparities in real estate appraisals. The second study, conducted months later by eAppraiseIT, argued that the disparity was mainly attributed to the type of dwelling being appraised rather than any form of implicit bias.
At the center of this disagreement is how each study obtained its data. CoreLogic’s data used mostly public records from county-level assessors, which meant the data set contained limited information regarding home features, ownership, and sale characteristics. To compensate for this, CoreLogic applied a hedonic pricing model in order to control for neighborhood characteristics. In contrast, eAppraiseIT provided a much larger data set since they had access to more detailed information from nearly 334,000 individual appraisals across the US.
The primary source of contention between the two studies is the way that geographic area was measured. CoreLogic did not specify how the geographic area was defined, whereas eAppraiseIT controlled for submarkets. This helped to explain why each study yielded different results—the difference in geographic areas allowed eAppraiseIt to capture more relevant information.
The disagreement between these studies highlights the importance of considering the context when interpreting data on appraisals. The CoreLogic study used an hedonic pricing model which accounted for neighborhood characteristics, whereas the eAppraiseIT study relied on controlling for small-scale units, such as submarkets. Both approaches can provide useful insights into the issue appraisal bias, but there is still much work to be done in order to fully understand and address this important issue.
This article provides insight into the debate surrounding appraisal bias in the mortgage industry. Researchers disagree as to whether or not implicit bias exists, with one study finding evidence for its existence while the other argues that any disparities are driven by the type of dwelling being appraised. At the heart of this disagreement is the type of data that was used in each study—CoreLogic relied on public county records while eAppraiseIT had access to over 300,000 individual appraisals. The differing results of the two studies highlights the complexity in understanding appraisal bias and suggests that further research is needed to draw definitive conclusions.
This article was contributed on Oct 19, 2023