Harvard Study: Home Building and Remodeling Cyclical

By: Jann Swanson

Calling it “an unusually cyclical industry…with frequently alternating periods of significant growth and decline,” The Harvard Joint Center for Housing Studies recently released a report on residential building cycles and the correlation between home construction and remodeling.

Understanding Remodeling Cycles by Abbe Will examines patterns of activity in homebuilding and remodeling, comparing them with overall economic activity over a 20 year period stretching from the first quarter of 1985 to the fourth quarter of 2007.  While this time frame includes only two recessionary periods (the current downturn in construction and other spending is not yet considered a recession), it was picked because of a change in research methodology by the Census Bureau in 1984.   

The housing industry is a major component of the U.S. economy; residential fixed investment totaled $641 billion in 2007.  Investment in new construction and residential improvements made up about 5 percent of gross domestic product (GDP) and 30 percent of private domestic investment on average since 1950. 

During this time period the industry experienced frequent booms and busts and the resulting impact of housing on the larger economy has generated much interest in studying the cycles and the reasons for them.  It is generally accepted that housing and the GDP tend to expand and contract during the same time periods and there is evidence that residential investment was the largest contributor to weakness in the GDP prior to eight of the past ten recessions.

These cycles, however, are not regular and predictable.  Reductions in spending for residential construction vary widely in both size and duration but generally lead a slowdown in spending on consumer durables.

Will notes that it is easy enough to see when the housing market is expanding or perhaps even overheating as home prices rise, the supply of homes shrinks, and homebuyers may find themselves in competitive bidding situations.   It is harder to anticipate when the market will cool and whether the bubble will burst or gently deflate.  How, in terms of time, do homebuilding cycles lead or lag remodeling cycles?  The study also seeks to deconstruct remodeling spending into components such as professional vs. do-it-yourself projects or improvement v. maintenance spending in order to identify the underlying drivers of patterns in the industry.

In deconstructing remodeling by project type, Will found that better than one-quarter (27.3 percent) of expenditures were made for interior and exterior replacements defined as roofing, siding, window and door installation, flooring and ceiling projects.  This category was followed (at 22.0 percent) by room additions and alterations including decks and porches as well as miscellaneous interior improvements.  Kitchens and baths represent 18.6 percent and improvements including adding or replacing a garage, fencing, septic systems claim 16.4 percent of the home improvement dollar.  Smaller amounts of money are spent on replacement of existing improvements such as plumbing repairs, electrical, HVAC and appliance expenses and for disaster recovery (10.4 and 5.3 percent respectively.)

Will says that, with so much segmentation the various components behave differently during the cycle.  Improvements, for example, are usually more discretionary than routine maintenance and repairs so the former is expected to be much more cyclical and sensitive to outside economic forces than the latter which is expected to be more stable.  High-end discretionary projects are more sensitive to changes in the remodeling cycle than the more routine and less costly improvement/repair component. 

In conducting the study Will used data from the Census Bureau’s C-50 series.  This is data aggregated quarterly and an annual rate of change computed on the total money spent over four consecutive quarters compared to expenditures in the four quarters preceding them.  The peak and the trough of the cycle are the points at which activity changes to slower growth followed by diminishing activity and vice-versa.

Will identifies three distinct remodeling cycles over the past twenty years.  The first began near the time of the 1990-91 recession and ended about five years later; the next cycle then started, lasted about three years, and ended near the time the 2001 recession began.  At 34 quarters thus far, the current cycle “is already about twice as long as the previous two,” and during this time remodeling spending grew to a maximum year-over-year growth rate of 20 percent (2005).  The greatest annual decline (-8.2 percent) occurred in the second cycle in 1999.

According to the author, there were several unique factors that led to the size and length of the current cycle.  The strong housing market and attendant appreciation in home prices contributed to a growing pool of home equity which, coupled with low interest rates and easy access to credit, led homeowners to cash out funds for home improvements.  The radical changes in two of these factors – falling house prices and tightening credit – are now forcing homeowners to retrench, curbing home improvement spending, and there is no indication that the current cycle is nearing the bottom.

When the annual rates of change in remodeling expenditures and new single family construction spending are compared, it is clear that both are very cyclical and follow similar patterns of increases and decreases.  The cycles over the last 20 years have been of similar magnitude but new home construction tends to peak a little higher and then fall to lower levels than remodeling cycles.  Changes in remodeling over the 20 year period have ranged between -8 percent and 22 percent while new construction spending moved in a larger range between -27 percent and 24 percent.  The four-quarter change rates averaged 5.9 percent for remodeling and 7.0 percent for new construction.

In terms of time, growth and decline in the two segments generally coincide.  In 82 percent of the quarters studied new construction and remodeling both experienced growth or decline at approximately the same time.  The remodeling cycles usually lasted longer than homebuilding cycles but the period of downturn (from peak to trough) tended to be shorter.  Remodeling downturns also tended to be less volatile than those for homebuilding.  At the peak of the last three cycles remodeling evidenced an annual rate of change of 22, 11.8, and 20.4 percent while for the same cycles new home construction had peak changes of 22.7, 16.7, and 23.8 percent.  At the bottom of each cycle the change for remodeling were -8.0, -8.2, and -5.7 percent while for new construction changes at the bottom of the cycle was -5.4, 2.3, and -27.1 percent.  (In the case of the current downturn the measurement was taken at the end of the fourth quarter of 2007 which, as we have subsequently learned, was nowhere near the bottom of the cycle.)

Breaking remodeling into its various segments identified earlier and using data from the American Housing Survey maintenance and repair spending was found to have grown steadily over time while the growth in spending for home improvements such as additions has varied over a wide range of 2.3 percent change to 20 percent.;

A high percentage of total improvement spending on single family homes goes to professional projects, an average of 86 percent per quarter from 1989 to 2007 with growth averaging 6.4 percent annually.  Do-it-yourself spending grew at an average rate of 4.7 percent year-over-year.

Spending for upper end discretionary projects is very cyclical while spending for systems and equipment replacements are more stable. 

The study concludes that owner improvement and repair activity tends to move in concert with both new residential construction and the general business cycle but is less volatile than homebuilding activity. Improvement spending is more affected by changes in the cycle than maintenance and repair spending and high-end discretionary spending is most sensitive to the business cycle.