Department of Numbers

The WSJ Offers Some Confusing Home Affordability Numbers

Posted Sunday, February 27 2011

I don't want to call these conclusions wrong because I have great respect for both the WSJ and Moody's Analytics (and Mark Zandi in particular), but the evidence provided in the articles Home Affordability Returns to Pre-Bubble Levels and Why 2011 May Be the End of the Housing Crash is very questionable.

Look at this claim made in the latter article for instance:

Housing is the most affordable it has been in decades, according to analysts at Moody's Analytics. They don't just look at house prices. They also look at incomes.

Nationally, the cost of a house is the equivalent of about 19 months of total pay for an average family, the lowest level in 35 years.

This is a pretty astounding claim and one that I'll suggest is at least misleading. First off, "19 months of total pay" translates into a price-to-income ratio of 1.58. On an absolute basis, that is a very low number (lower numbers implying greater affordability) and suggests that the typical home costs only 1.58 times what the average family earns annually. If we're trying to get at what's typical, does that sound right to anybody? The question to ask then is how they arrived at that number. Since I don't have the Moody's report and since the author of the piece hasn't said, all we've got to go on is the reference in the sentence "the cost of a house is the equivalent of about 19 months of total pay for an average family."

Home prices are notoriously hard to track, so the time series you use to measure them is important. Similarly, income can vary widely depending on what demographic you measure and what aggregate statistic you choose. I can't even speculate on the home price series that went into this analysis, but the income data appears to be for the average family (although what survey it's from is not clear). Average family income skews high (i.e. average income does not represent the middle of the income distribution), and it disregards households that are not part of families (which typically don't have as many wage earners). Why would they choose to report based on this metric? Did they then compare average household or family income to a median price? What was the rationale for this? I suspect Moody's has a good reason for choosing these measures, but that reasoning doesn't appear to have made it into either article.

Though there's no perfect way to measure home affordability, median income and median home prices provide a much better measure of the middle than averages do. So let's look at the price-to-income ratio using those figures. The median sales price in Q4 of 2010, according to NAR's sales figures for existing homes, was $170,600. The Census CPS shows the median income for a household was $49,777 in 2009, the latest data available. Using those two numbers we can easily calculate a price-to-income ratio of 3.43 — more than twice what the WSJ reported. I'm not sure where they got the average family income data from, but using the NAR median price implies the average family income they used was roughly $107,000 — quite high and certainly beyond typical.

Using FHFA price statistics we can see the price-to-income ratio back to 2000 (expressed in dollar-to-dollar terms). The current Realtor ratio I calculated above is marked in red. As you can see, these two home price series compare favorably.

Price to Median Household Income Using: FHFA Home Prices, NAR Home Prices

The other claim that I take issue with is that the price-to-income ratio has not been lower for 35 years. Maybe when you use the price-to-income ratio as they've constructed it, this is true. When you instead use median household income with a long term index from Case-Shiller or the FHFA, the ratios were definitely lower in the late 90s up to 2000 when they began to take off. Today the price-to-income ratio is about 20% higher than 1998. There's no doubt that things are significantly more affordable today than they were 5 years ago. The claim that they are the most affordable in a generation seems to be going too far however.

Price to Median Household Income Index Using: FHFA Home Prices, Case Shiller Home Prices, Core Logic/First American Home Prices

Note: The chart above expresses the ratio as an index set to equal 100 in 2000.

The tone of both of these articles reminds me of one of the elements of the bubble that we don't talk about much these days. During the 2003-2005 time period there was a real feeling that you were perhaps witnessing the only opportunity you might ever have to own a home. A lot of people who bought back then were called greedy because they wanted to own an asset that would appreciate at 10-15% annually forever. I've always believed that a huge number of people bought because they asked themselves if they'd ever be able to afford a home in the future if they didn't buy then. With prices appreciating so fast, homes were bound to be out of reach forever sooner or later — or at least that was the logic. The way these articles talk about real estate reminds me of that kind of "get in while you still can" mentality that reigned 10 years ago and supported that anxiety.

In the end I don't dispute the point that homes are considerably more affordable than they have been or that it makes more sense to buy a home you can afford today than it has for some time. Perhaps this all just exposes my preference that affordability analysis not get mixed up with speculative assessments of price direction. But I could easily forgive that if it was just a bit clearer how and why particular data sources were used to provide evidence for these claims. What is very clear to me is that you can't just state a price-to-income ratio (or any affordability ratio for that matter) without also defining a methodology.