SMU Construction Analysis: Nonres, Infrastructure Lead in Spending

At SMU, we analyze the CPIP data with the intent to provide a clear description of activity that accounts for about 45 percent of total U.S. steel consumption. See the end of this report for more detail on how we perform this analysis and structure the data. In particular, note that we present non-seasonally adjusted numbers and we don’t include expenditures for residential improvements. Our rationale is that construction is highly seasonal and our businesses function in a seasonal world, and home improvements don’t consume much steel. The growth or contraction that we report in this analysis has had seasonality removed by providing only year-over-year comparisons. For these reasons, our results may not align with other sources.

Total Construction

Total construction expanded by 6.6 percent in the three months through September and by 4.2 percent in the 12 months through September, both year over year. Since the three-month growth rate is higher than the 12-month rate, we conclude that momentum is positive. September construction expenditures totaled $101.9 billion, which breaks down to $71.6 billion of private work, $28.1 billion of state and locally funded (S&L) work, and $2.2 billion of federally funded work (Table 1). Growth trend columns in all four tables in this report show momentum.

Figure 1 shows total construction expenditures on a rolling 12-month basis as the blue line and the rolling three-month year-over-year growth rate as the brown bars. Total construction is at an all-time high and growth has accelerated throughout 2018.

Figures 1, 2, 6, 7 and 9 in this analysis have the same format, the result of which is to smooth out variation and eliminate seasonality. We consider four sectors within total construction. These are nonresidential, residential, infrastructure and other. The latter is a catchall and includes industrial, utilities and power. Of these four sectors, only residential buildings had negative momentum in the September data, which was the same as the previous eight months’ results.

The pre-recession peak of total construction on a rolling 12-month basis was $1.028 trillion through November 2006. The low point was $665.1 billion in the 12 months through April 2011. In 12 months through September 2018, construction expenditures totaled $1.092 trillion. (This number excludes residential improvements; see explanation below.)

Private Construction

Table 2 shows the breakdown of private expenditures into residential and nonresidential and subsectors of both. The growth rate of private construction in three months through September 2018 was 5.2 percent, up from 1.9 percent in the three months through last November, as shown by the brown bars in Figure 2. May was the first month for the rolling 12-month total of private construction expenditures to exceed the pre-recession peak that occurred in September 2006. Private construction expenditures have increased every month since September 2011 on a rolling 12-months basis.

Within private non-residential buildings, momentum was mixed with seven positive sectors and two (manufacturing and religious) negative. The weakest category is religious buildings, which have had negative year-over-year growth every month since March 2017. 

Excluding property improvements, our report shows that single-family residential grew by 4.4 percent in three months through September with negative momentum and multifamily residential grew by 2.9 percent with positive momentum. The higher federal funds rate is starting to show its effects in the housing sector. Housing starts, existing and new home sales, as well as other housing-related data, have been mostly disappointing in recent months. This is predominantly due to the substantial rise in mortgage rates, a direct effect of a higher funds rate. In fact, the MBA 30-year mortgage has risen to 5.05 percent, its first time above the 5 percent point since 2011. Also, the single-family market is experiencing headwinds from the effects of tax reform. Homebuilder sentiment has been declining since February, but is still higher than the financial conditions say it should be.

The Census Bureau reports on construction starts in its housing analysis. In the starts data, the whole project is entered into the database when ground is broken. Construction put in place is based on spending work as it occurs; the value of a project is spread out from the project’s start to its completion. Single-family starts grew by 2.5 percent in the three months through September year over year, which was close to but a bit lower than the growth rate of CPIP. Multifamily starts grew by 0.4 percent, which was quite a bit lower than CPIP. Figure 3 shows the growth of both housing sectors since January 2005 and Figure 4 shows the ratio of single-family to multi-family starts. The proportion of single family has been gradually increasing since September 2015 when the ratio was 1.66 in favor of single units to 2.58 in September 2018. Figure 5 shows total housing starts in four regions with the South being the strongest and the Northeast the weakest.

State and Local Construction

S&L work has expanded and accelerated every month this year. In September, the year-over-year growth rate on a rolling three-months basis reached 10.4 percent, up from 0.9 percent in January and negative growth every month in 2017 (Table 3). S&L nonresidential buildings and infrastructure, both of which are included in the S&L data, had positive momentum. Table 3 shows that S&L nonresidential buildings expanded by 11.1 percent in three months through September. Educational is by far the largest subsector of S&L nonresidential buildings at $6.6 billion in September and experienced a 4.2 percent positive growth on a rolling three-months basis with positive momentum. Figure 6 shows the history of total S&L expenditures. In previous updates we have stated that a full recovery of S&L construction wouldn’t happen in this decade. The growth of this sector has exceeded our expectations in the last three months and is now so great that this conclusion may no longer be valid.  

Drilling down into the private and S&L sectors as presented in Tables 2 and 3 shows which project types should be targeted for steel sales and which should be avoided. There are also regional differences to be considered, data for which is not available from the Commerce Department.

Infrastructure

Infrastructure expenditures peaked in early 2016 and declined through 2017 before beginning a robust recovery in 2018. September had a 9.9 percent growth rate. Highways and streets including pavement and bridges account for about two-thirds of total infrastructure expenditures. Highway pavement is the main subcomponent of highways and streets and had a 10.1 percent growth in three months through September. Bridge expenditures achieved positive growth last December for the first time in almost two years and steadily improved until this latest data for September when there was a slight decline (Table 4).

Figure 7 shows the history of infrastructure expenditures and the year-over-year growth rate.

Total Building Construction Including Residential

Figure 8 compares year-to-date expenditures for building construction for 2017 and 2018. Single-family residential is dominant and in the 12 months of 2017 totaled $264.5 billion, up from $242.5 billion in 2016. The rate through the first nine months of 2018 was $285 billion. It seems likely that the 2018 tax bill, by its reduction in the deductibility of mortgage interest and local taxes, will negatively affect single-family home construction.

Figure 9 shows total expenditures and growth of nonresidential building construction. Growth slowed in January through September last year, went negative in September through October, then became positive in November through September 2018. On a rolling 12-months basis, expenditures on nonresidential buildings including both private and S&L were at an all-time high exceeding the previous best set in December 2008.  

Explanation: Each month, the Commerce Department issues its construction put in place (CPIP) data, usually on the first working day covering activity one month and one day earlier. There are three major categories based on funding source: private, state and local, and federal. Within these three groups are about 120 sub-categories of construction projects. At SMU, we slice and dice the expenditures from the three funding categories to provide as concise a summary as possible of steel consuming sectors. For example, we combine all three to reach a total of nonresidential building expenditures.  Construction put in place is based on spending work as it occurs, estimated for a given month from a sample of projects. In effect, the value of a project is spread out from the project’s start to its completion. This is different from the starts data published by the Census Bureau for residential construction, by Dodge Data & Analytics and Reed Construction for nonresidential, and Industrial Information Resources for industrial construction. In the case of starts data, the whole project is entered to the database when ground is broken. The result is that the starts data can be very spiky, which is not the case with CPIP.

The official CPIP press release gives no appreciation of trends on a historical basis and merely compares the current month with the previous one on a seasonally adjusted basis. The background data is provided as both seasonally adjusted and non-adjusted. The detail is hidden in the published tables, which SMU tracks and dissects to provide a long-term perspective. Our intent is to provide a route map for those subscribers who are dependent on this industry to “follow the money.” This is a very broad and complex subject, therefore to make this monthly write-up more comprehensible, we are keeping the information format as consistent as possible. In our opinion, the absolute value of the dollar expenditures presented are of little interest. What we are after is the magnitude of growth or contraction of the various sectors. In the SMU analysis, we consider only the non-seasonally adjusted data. We eliminate seasonal effects by comparing rolling three-month expenditures year over year. CPIP data also includes the category of residential improvements, which we have removed from our analysis because such expenditures are minor consumers of steel.

In the four tables included in this analysis, we present the non-seasonally adjusted expenditures for the most recent data release. Growth rates presented are all year over year and are the rate for the single month’s result, the rolling three months and the rolling 12 months. We ignore the single-month year-over-year result in our write-ups because these numbers are preliminary and can contain too much noise. The growth trend columns indicate momentum. If the rolling three-month growth rate is stronger than the rolling 12 months, we define that as positive momentum, or vice versa. In the text, when we refer to growth rate, we are describing the rolling three-month year-over-year rate. In Figures 1, 2, 6, 7 and 9 the blue lines represent the rolling 12-month expenditures and the brown bars represent the rolling three-month year-over-year growth rates.