New Construction Starts is powerful data if used properly. Understand how to use the data and you have an excellent forecasting tool.
New Construction Starts is incorrectly used when the data is referred to as construction spending. It cannot be used to look at the year over year (yr/yr) or month over month (mo/mo) trend in values to predict % change in construction spending. This misrepresents how to use New Starts data.
Care must be taken to use Starts properly. It is sometimes misinterpreted in common industry forecasting articles. Starts dollar values represent a survey of about 50% to 60% of industry activity, therefore Starts dollar values cannot ever be used directly to indicate spending. Also, Starts do not directly indicate changes in spending per month or per year. Projected starts data cannot be used to directly forecast expected construction volume.
Only by including an expected duration for all Starts and producing a forecast Cash Flow from Starts data can the expected pattern of spending be developed. Finally, it is the rate of change in Starts Cash Flows that gives an indication of the rate of change in spending.
Cash flow is the best indicator of how much and when spending will occur. Cash flow from starts gives a prediction over time of how spending from each month of previous starts will change from all projects in backlog. Cash flow totals of all jobs can vary considerably from month to month, are not only driven by new jobs starting but also old jobs ending, and are heavily dependent on the type, size and duration of jobs.
New Starts for the month or the year is the total value of new project revenues that came under contract in that period. Since the values reported for Starts are a sampling survey of about 60% of the industry totals, the total dollar volume is not comparable to actual spending. However, the percent change in values is very useful.
The entire value of a project enters backlog when the contract is signed and work begins. That’s a new start. Projects booked on or before December 31st, that still have work remaining to be completed, are in backlog at the start of a new year.
Simply referencing total new starts or backlog in the year does not give an indication of spending within the year or next calendar year. Projects, from start to completion, can have significantly different duration. Whereas a residential project may have a duration of 6 to 12 months, an office building could have a duration of 18 to 24 months and a billion-dollar infrastructure project could have a duration of 3 to 4 years. New starts within any given year could contribute spending spread out over several years. Total Cash Flow in the Year, or Spending, could include spending from projects that started years ago.
Backlog at the start of the year could include revenues from projects that started in December or several years ago. For a project that has a duration of several years, the amount in starting backlog at the beginning of the year is the amount remaining to complete the project or the estimate to complete (ETC). And all that ETC may not be spent in the year following when it started, dependent on the duration remaining to completion.
The only way to know how much of total starts or total backlog that will get spent in the current year and following years is to prepare an estimated cash flow from start to finish for all the projects. The sum of the amounts from all projects ongoing in each month gives total cash flow in that month, or monthly spending in that year. Spending in any given month could have input from projects that started many months ago. The sum of the cash flow is what shows the expected change in spending.
The following table clearly shows there is not a correlation between starts in any year with spending in the following year. The practice of using construction starts directly to predict spending in the following year can be very misleading in an industry that relies on data for predictive analysis to plan the future. Not only does it not predict the volume of spending in the following year, it does not even consistently predict the direction spending will take, up or down, in the following year. It’s a false indicator and it’s not a good use of data.
Construction spending is strongly influenced by the pattern of continuing or ending cash flows from the previous two to three years of construction starts.
Current month/month, year/year or year-to-date trends in starts often do not indicate the immediate trend in spending.
Power market starts and spending provides a good example.
Power starts peaked in 2015 at an all-time high, up 140% from 2014 and more than the prior two years combined. Yet Power spending was down 6.5% in 2015 and down 1.5% in 2016. This happened because Power starts were also at an all-time high in 2012, just below the 2015 level, and those starts drove 2014 spending to an all-time high, but then tapered off in 2015. Those peak starts from 2015 will still be contributing spending for several years to come, long beyond typical jobs, and that drives typical spending growth because it adds more than typical number of months that contribute spending.
Power starts gained only 1.5% in 2016, dropped 7% in 2017 and are expected to finish 2018 down 12%. The pattern of cash flows from starts is indicating growth in spending for 2018 and 2019. Starts from 2015 and 2016 with longer than average duration contribute spending out to 2020 & 2021, breaking the average balanced cycle of one month of old jobs ending for every new month of jobs starting. That drives the pattern in spending.
The following example shows what happens to monthly spending growth when a long duration job first influences spending past the typical duration and then when it ends. In the example table presented below, starts grow at 1% per month and have a typical duration of 5 months. But one unique month has an unusually large project start that will last for 10 months.
A typical month of spending has cash flow from 5 months of starts, but the long duration project creates 6 months of cash flows for the period beyond typical duration. Notice what happens and when it occurs.
When the large project starts it has no unusual effect on spending. But when it first extends beyond typical duration, it has a massive +20% growth effect on spending, even though starts had only been increasing at 1%/month for the previous 5 months. When it ends it has a similar downward effect, again, even though starts had been increasing at 1%/month.
Spending growth (or declines), both when an extra-large job causes it to increase and then when the extra-large job ends, is almost entirely influenced by the long duration project, not by normal monthly starts growth rate. This same example can be over months or over years.
Spending patterns are far more influenced by projects with unusual duration. Construction spending is strongly influenced by the pattern of continuing or ending cash flows from the previous two to three years of construction starts. Cash flow is the best indicator of how much and when spending will occur.
All construction starts data in this report references Dodge Data & Analytics Starts data.
U. S. Census posted Construction Spending for July at a seasonally adjusted annual rate (SAAR) of $1,315 billion, up only 0.1% from May.
Year-to-date, July construction spending is up 5.2% from the same period in 2017.
June was revised down slightly, -0.2%, and May was also revised down, -0.6%, but May remains up 1.7% from the 1st May release.
Construction Spending for the 1st 7 months of 2018, in Current $, by Census formulas averages $1,306 billion. By my formulas the 1st 7 months average stands at $1,321 billion. Either way, this is an all-time high, well above the pre-recession high spending of $1,205 billion posted in the 1st quarter of 2006. Spending has been above the 2006 high since the 4th quarter 2016, but since 2006, no other 6-month period has averaged above $1,250 billion. Spending is expected to total $1,335 billion for 2018.
Constant $ shows volume reached peak during the 2nd half 2005 and 1st half 2006, with 2005 posting the peak year. 2018 constant $ inflation adjusted spending is still 14% below the 2005-2006 peak.
Total spending year to date through June is $740 billion. Historically, 56% of annual spending occurs in the 1st 7 months. Jan, Feb and Mar are the weakest months of the year, while Jul, Aug and Sep are the strongest spending months. This would indicate a 2018 total annual spending of $1,321 billion, 1% less than my forecast.
Top performing construction spending markets 2018 year-to-date through July are Transportation +15.8%,Water Supply +14.1%, Public Safety +13.1%, Conservation 10.3%, Lodging +10.1%, Sewage and Waste Disposal +9.1%, Residential +7.6% and Office 7.2%.
The only markets down year-to-date are Religious -11.8% and Manufacturing -7.5%. Religious building as a percent of total is so small (1/4 of 1%) it has negligible effect on total annual performance. However, Manufacturing is about 5% of total construction.
Residential, Office, Commercial/Retail, Lodging, Highway and Environmental Public Works (Sewage, Water, Conservation) are all ahead of my expectations for the 1st half of 2018.
Last month, June construction spending showed an unusual $9 billion (SAAR) monthly decline (-9.3%) in Educational spending. At that time I said, “This is several billion greater than the largest decline reported during the recession, so this looks like an anomaly in the data. There has never been a monthly decline like this in the Educational market since I’ve been tracking data, back to 2001. It is double the largest non-recession decline. I expect it will be revised up substantially at some point in the future.” That anomaly in the June data was revised up this month, erasing about half of the decline that was first reported.
Transportation is another market that appeared to be unusually low for June. Last month I said this, “Transportation (terminals and rail) new starts in 2016 increased 34% and then in 2017 increased 120%. Even with long duration cash flow spreading out the spending for big projects, my analysis still predicts Transportation spending up 30% in 2018. Year-to-date through June, Transportation spending is up only 14%. I’ve forecast it should be up 18%. That’s a total shortfall of about $1 billion (SAAR ~$12 billion), or about 7%/month, for 3 months. April, May and June spending are all below expectations.” In the July data, both May and June spending were revised UP by a total of $2 billion. With that revision Transportation spending is up 18% YTD through June, as expected.
Manufacturing spending as of June was reported down 8.7% year-to-date from 2017. Spending through July is now down only 7.5%. I previously reported that I expect the decline to slowly turn positive in the second half of the year to finish up 2%. Spending is currently at an SAAR just above $66 billion and expected to increase to $70 billion by December. In 2017, spending started the year above $70 billion but decreased to $60 billion by year end. Increasing values in the 2nd half 2018 compared to decreasing values in 2017 will continually increase the year-to-date performance in the 2nd half of 2018.
Power, similar to manufacturing, posted the highest spending for 2017 early in the year, then declined. In 2018, the 1st half posted the lowest spending, so the year-to-date is currently low. Increased spending in the 2nd half 2018, compared to the lowest values of the year in 2017, will boost year-to-date spending every month through year end. Although year-to-date spending through July is up only 0.7%, I expect the total for the year will finish up 8%.
Manufacturing and Power highlight one of the biggest shortfalls of judging expected performance based on year-to-date change. It is important to look at the trend line expected in the current year versus the trend line in the previous year. If they diverge, then year-to-date change will not give a clear indication of expected performance in the current year. Manufacturing data as an example follows. Note, SAAR data shows performance trend but NSA$ is needed to get YTD$.
Public spending increased 5% in 2015, but has been depressed since since 2009. 2017 finished still 7% lower than 2009. For 2018 we should see a gain of $16 billion, +5.7% over 2017 to $308 billion, the highest finish since 2009. Highway and Street is the largest share of public work, but adds very little to 2018 gains. Educational spending makes up about 25% of all public spending gains. Public Works (Sewage/Waste Water, Water Supply and Conservation), only 14% of all public spending, accounts for about 25% of the gains this year. Public Transportation, at only 12% of public spending, accounts for $8 billion in increases in public spending, half of all the gains in public spending this year.
Total spending has increased from an average of $1,254 billion in Q4’17 to $1,292 billion in Q1’18 to $1,321 billion in Q2’18, growth of 3.0% and 2.25% the last two quarters. I’m expecting the rate of monthly spending will be above $1,360 billion by year end. The total spending forecast for 2018 is $1,335 billion.
Residential single family spending is up 8.5% YTD. Multifamily is down 0.9%. Total residential spending is forecast to reach $570 billion in 2018, growth of 7.2% over 2017.
Nonresidential Buildings spending YTD totals $246 billion, up only 1.7% from 2017. It is being held down by Manufacturing which is currently down 7.5% from 2017. 2018 forecast is $445 billion, 6.1% growth over 2017, with best growth in Lodging 13%, Office 11% and Amusement/Recreation 9%.
Non-building Infrastructure will post the best year of growth since 2014 to reach a new all-time high at $308 billion. Transportation, by far, will show the best growth, 25% above 2017.
Cash flow from backlog supports a 2018 spending forecast of $1,335 billion, a spending increase of 7.2% over 2017. The forecast for 2019, based on a modest 3% increase in new starts in 2019 is $1,400 billion, an increase of 5% over 2018. The strongest growth in spending for 2018 and 2019 is forecast to occur in Non-building Infrastructure with Transportation being by far the strongest market.