Single-Family and Multifamily Starts: Long-Run Trends


Growth in multifamily construction is leading the recent increase in housing starts. And with many analysts focusing on multifamily development, we thought it would be useful to consider the long-run relationship between single-family and multifamily housing starts.

Over the last 40 years, a discernible trend has emerged with respect to the ratio of single-family to multifamily housing starts. Overall, both types of starts have been subject to considerable variation due to business cycle events and/or housing policy changes.

For example, multifamily starts were lifted in the early 1970’s thanks to the promotion of the Section 8 program, but in turn fell dramatically once the Baby Boom generation entered its prime homeownership years. After the Tax Reform Act of 1986, which increased the cost of ownership of multifamily rental properties, multifamily starts again declined.

Despite these ups and downs, a long-term trend is present in the ratio of single-family to multifamily starts. In the graph below, the blue line plots this ratio, with higher values corresponding to larger multiples of single-family starts compared to multifamily starts. The black line plots the long-run linear trend, which clearly has been increasing. Several reasons explain this particular trend.

One such explanation is plotted on the graph: the aging of the United States. The red line below tracks the median age of the U.S. population, which has been growing as the size of the Baby Boomer cohort overwhelms the smaller numbers found within Generations X and Y. An aging population is one factor that will yield a higher homeownership rate over the long-term and because single-family housing units are the preferred structures of homeowners, this demographic effect has increased the ratio of single-family to multifamily starts.

However, an aging population, as measured by the median, does not reveal all the information about the distribution of various age cohorts. So another proposed explanation for the differences in single-family and multifamily construction is the share of the U.S. population in their 20’s. In general, individuals in their 20’s have a higher propensity to rent, as they build savings for a down payment for a home purchase. Therefore, since renters are more likely to live in multifamily apartment buildings, a larger share of the population in their 20’s should mean a higher level of multifamily development. This is graphed as the red line below (the share of the U.S. population above the age of 15 that are in their 20’s).

Putting these possible explanations together, we estimated a simple regression model to determine the specific impacts. The results indicate that a 1 percentage point increase in the share of the population in their 20’s decreases the single-family /multifamily starts ratio by 0.17. As the ratio currently stands at 2.33 (for October 2011), this would represent a more than seven percent decrease.

The regression model results also indicate that an increase of 1 year for the median age of the U.S. would increase the starts ratio in favor of single-family construction by 0.2 or about 9%.

What does this tell us about the future? Without a doubt, the median age of the U.S. will increase, and as aging-in-place becomes more common, the demand for single-family starts will be supported by an aging population. However, as the remainder of the Echo Boom generation (born 1982 to 1995) enters their 20’s, this will in turn support multifamily development. And in the short-run, housing policy changes, such as those that would require or necessitate larger down payments to purchase a home, mean that multifamily rental demand will be strong.

Taken together, this suggests that the ratio of single-family to multifamily starts in home building will fall in the short-run (e.g. over the next two years) before returning to its upward trend over the long-run.


Tags: , , , , ,

Leave a Reply

Your email address will not be published. Required fields are marked *

%d bloggers like this: