Compared to the Case-Shiller index alone this is a useful analysis, but it doesn't take into account that most homes are bought using mortgages, not cash, so the cost of money must be factored in.
It is easy enough to access historical mortgage rates to see the rates people paid to buy homes over this period:
The raw WSJ index data behind their snazzy chart can be gleaned from
Using a Groovy script I parsed their index data and used a multiplier based on the 30-year mortgage rate. As was done in the article, the resulting data was normalized to the first point in Jan '87 so the graphs would be comparable.
Using my "True Cost" index, the summer of 2006 was still very elevated (my index of 150 compared to 100 in Jan '87) but not quite as crazy as it would appear by not taking into account the prevailing mortgage rates (WSJ index 200 compared to 100 in Jan '87).
March of 1989, with its 30-year rate at 11.03%, was also a crappy time to buy a house. The true cost of housing actually did not return to the '89 level until the height of the boom in '06. Refinancing would only offer some comfort to the '89ers- you would have had to wait until August of 1992 to refinance below 8% and until Jan of 2003 to get below 6%.
With today's insanely low rates we see my index is at a manageable 84, similar to May of 1999, instead of the alarming 128 Arends has used as a metric for his article. A dip much below 5% would mean the cheapest mortgaged houses relative to earnings of in over 20 years.