The story of household formation weakness is about families

Traditionally analysts tend to look at total households in the US as an indicator of demand for housing.  Many have pointed out that recent years generated below average household formation. For example this presentation from Harvest Partners puts together a clear argument for pent up demand in housing based on analysis of recent poor growth in total households.

The US Census however splits households into two categories: "family" and "non-family". About 66% of US households are families, with the rest classified as "unrelated people" sharing a living space - college students, unmarried couples, singles living on their own, etc.  The chart below (last 20 years) shows that non-family households have generally been growing in line with the US population and although dipped in 2008, have since recovered.

Non-family households vs. the US population (thousands, source: US Census Bureau)

The real problem however is found in the family household formation. Family households have completely decoupled from the US population growth since 2008.

Family households vs. the US population (thousands, source: US Census Bureau)

This deviation is quite new.  Family households have been forming at an average rate of 651,000 per year since at least 1947 (when the first annual household data became available). During that whole period the only years showing "negative formation" are 2008, 2010, and 2011 - likely the result of families moving together (parents and grandparents, etc.).

Clearly the simplest explanation for this divergence is the spike in unemployment and economic uncertainty. People are forgoing or delaying forming separate family households (such as delaying marriage). As JPMorgan pointed out last week, this should correct itself if employment improves.

Jamie Dimon (JPM): We've added 10 million Americans. We're going to add 3 million Americans every year for the next 10 years, that's 30 million Americans who need 13 million in drawings or something like that. Household formation has been half what it normally is, and in most economy tell you that's going to come back with job creation.

But is family household formation caused by the recent poor economic conditions in the US or is it simply a correlation?  Most would say this causality is obvious, but is there evidence to support it?  Could there another explanation?  As an example of mistaking correlation for causality consider the age old belief  that poor economic conditions and unemployment cause an increase in violent crime rates. Apparently there is no hard evidence to support this view.  As Michael Shermer pointed out, "It's NOT the economy, stupid".

Understanding the true reasons behind this sudden stop in family household formation is critical to answering the question about pent up demand for housing that will ultimately drive the US housing recovery. But whatever the reason, analysts should be focused on families instead of the overall household number for signs of improvements in household formation.

SoberLook.com