In addition, housing is an important source of collateral for household borrowing, and, as we have seen, its value can be subject to considerable fluctuation. Also, research has shown that having a mortgage that is larger than the value of the underlying house is associated with an elevated risk of default.
Measuring aggregate house prices, estimated using house price indexes (HPIs), tends to be surprisingly difficult. Furthermore, the methodologies that are most accurate in the long run are subject to considerable revision in the short run. For example, monthly revisions to a commonly used HPI for the United States released by CoreLogic tend to be large up to about the third revision. There is also some evidence that these revisions tend to be downward, although there is currently no clear theoretical explanation for this phenomenon. Therefore, HPI users who are interested in the most recent data should interpret them with caution.
In this report, I review the literature on HPI methodologies, revisions, and sources of bias. I first describe the methodology common to all “repeat-sales” indexes, some improvements on that methodology, and some competing approaches. I then discuss some sources and characteristics of revisions to repeat-sales indexes.