Earlier today I drew attention to the faulty Demographia reports that the media have lapped up. Larry Mitchell emailed me about it:
Yes! … you have raised some significant issues of measurement, largely those related to income data concerned with the Demographia Housing affordability index.
I too have some major reservations principally concerned with the survey’s obdurate stance that ignores the influence of movements of incomes which affect Housing affordability.
I have tried, (on a number of occasions and unsuccessfully) to point out to its authors that the Demographia index/survey of Housing affordability, by ignoring income effects results in merely a good, but far from comprehensive “first cut indicator” of Housing affordability. To claim for the survey anything more rigorous than this, or to base affordable housing public policy on these findings moves toward the much more problematic.
Housing costs above all else in this context rely on many, mostly market-driven variable preferences, simply put, the matching of needs/wants with resources. From this complex situation the Housing Affordability index takes just two of the many variables, Housing costs and Incomes.
What is worse, the survey findings blithely ignore the impact (index movement-causation) of the income variable blaming all! of the Housing affordability woes on the other element … of Housing costs.This myopic focus severely limits the usefulness of the survey.
Lets briefly examine any presumption that the Income to House price ratio is a “reliable” indicator of housing affordability driven as is claimed by its authors by Housing costs alone.
If the ratio moves from (say) 3.0 to (say) 4.0 (less affordable) how is it possible to simplistically just attribute the ratio increase merely to the single Housing cost element? After all, there are two variables in the ratio, Both influence the measure but one is ignored in the analysis … that is, “Incomes”.
Housing costs are not the sole Housing affordability determinant obviously for if (real) Incomes have (say) declined at the same time as (say) housing costs have remained constant, then housing will be less affordable solely as a result of lower incomes. “QED”.
Using Demographia’s evaluations though, the index movement to lower affordability would (wrongly) be attributed to increased Housing costs (that had not changed!).
Until this Demographia measure accepts that some form of regression or other statistical weighting analysis is used to tease out the relative influences of BOTH factors upon the ratio we should treat the Demographia findings for what they are … simplistic “first cut indicators”.
Larry N. Mitchell