In 2015, nonresidential buildings starts were very high in the beginning of the year and dropped off in the later part of the year. In 2016 it’s just the opposite. This skews year-to-date total comparisons and for most of this year makes it appear as if there may be no growth in new starts.
Here’s a simple example:
Let’s say 4 months in 2016 had starts of $6, 8, 10 and 12 billion and the same months in 2015 had starts of $10, 9, 8 and 7 billion. The year-to-date change for 2016 vs 2015 after the 1st month (6 vs 10) is down 40%. After two months it’s 14 vs 19 (6+8 vs 10+9), down 26%. In the 3rd month 2016 has better performance than 2015 (10 vs 8), but the year-to-date (24 vs 27), down 11%, is still strongly influenced by the earlier months. But in the 4th month we get 36 vs 34 and finally the year-to-date shows 2016 growth of 6% over 2015. That is the current scenario.
Construction Starts for nonresidential buildings for the 1st 4 months in the 2nd half of 2016 (Jul-Oct) are 30% higher than the average of the 1st half 2016 and almost 40% higher than the same 4 months in 2015, and yet the year-to-date % change 2016 vs 2015 is ZERO.
To keep from being misdirected, year-to-date comparisons require knowing not only the direction of the current year trend but also the direction of the previous year trend.
The most recent 3 month (Aug-Sep-Oct) average of nonresidential buildings construction starts by Dodge Data represent the best 3 months since Q1 2008. Although year-to-date performance of zero growth would seem to indicate a slow down, starts are doing just fine. I’m forecasting the final two months of the year to be up 40% from 2015.
But wait, there’s more!
Every year, starts from the previous year are adjusted, always higher. 2016 starts won’t be adjusted up until 2017. But that means all current 2016 (un-adjusted) starts are being compared to 2015 that has already been adjusted up. This causes the year-to-date comparison to be always understated. The average adjustment to nonresidential building starts for the last few years has been about +5%. If that trend remains consistent then next year we should see that 2016 starts were approximately 5% higher than first posted and growth was really much better than current values would seem to indicate.
With 10 months of data in hand, year-to-date starts for nonresidential buildings show no change from 2015. However starts are doing very well and I’m predicting the 2016 volume of starts will lead to 8% growth in spending in 2017.
Why the big slow down in construction jobs this Year? Is work volume on the decline? Are labor shortages to blame?
These days, the most talked about reason for slower jobs growth is the lack of experienced workers available to hire. In fact, recent surveys indicate about 70% of construction firms report difficulty finding experienced workers to fill vacant positions and the Job Openings survey has been at highs for several months. That certainly cannot be overlooked as one reason for slower jobs growth, but that is not the only reason?
Although recent growth has slowed, even with all this talk of difficulty finding experienced construction workers, there has been very good jobs growth. For the 5 ½ year period from the low point in January 2011 to the present (August 2016) we added 1,240,000 construction jobs.
- Jobs increased by 23% in 5 ½ years.
- Spending growth increased 52% in that same 5 1/2 years.
Why is it that jobs don’t increase at the same rate as construction spending? Because much of that spending growth is inflation, not true volume growth. Volume is construction spending minus inflation. To get volume, convert all dollars from current $ in the year spent into constant $ by factoring out inflation.
- Spending growth is not a true measure of increase in real volume.
- Jobs growth should not be compared to spending growth.
Now that we have spending converted to volume, we need to adjust jobs to get real work output. The total hours worked affects the entire workforce so has a significant impact on output.
- Jobs is not a true measure of work output.
- Jobs x hours worked gives total work output.
Spending must be factored to remove inflation and jobs should be factored to include any change in hours worked.
- In the 5 1/2 years from Jan 2011 to mid 2016, real construction volume and jobs/hours real output both increased by 28%.
Now we see over the long term, job/hours and real volume are moving in tandem. But there are always short term periods when they do not and that causes ups and downs in productivity.
In 2014-2015, jobs/hours grew by 11%, the fastest growth for a two-year span in 10 years. Real volume of work increased by 16% producing a real net increase in productivity. But productivity had declined significantly in 2010 and 2011. It’s not unusual to see productivity balancing out over time. In part, this is due to companies balancing their total employees with their total workload.
From October 2015 through March 2016, jobs growth was exceptional. 214,000 construction jobs were added in 6 months, topping off the fastest 2 years of jobs growth in 10 years. That is the highest 6-month average growth rate in 10 years. That certainly doesn’t make it seem like there is a labor shortage. However, the jobs opening rate (JOLTS) is the highest it’s been in many years and that is a signal of difficulty in filling open positions.
- For the 6-month period including Oct’15 thru Mar’16 construction gained 214,000 jobs, the fastest rate of consecutive months jobs growth in 10 years. Then, after 3 months of job losses, July, September and October show modest gains.
I would expect growth such as we’ve had for two years and then that 6 month period to be followed by a slowdown in hiring as firms try to reach a jobs/workload balance. It appears we may have experienced that slowdown. Jobs declined for four of five months from April through August. Keep in mind, this immediately follows the fastest rate of jobs growth in 10 years. But it also tracks directly to three monthly declines in spending. (I predicted this jobs slowdown in my data 9 months ago. I predicted the 1st half 2016 spending decline more than a year ago).
It is not so unusual to see jobs growth slowed for several months. It follows directly with the Q2 trend in spending and it follows what might be considered a saturation period in jobs growth. The last two years of jobs growth was the best two-year period in 10 years. It might also be indicating that after a robust 6 month hiring period there are far fewer skilled workers still available for hire. The unemployed available for hire is the lowest in 16 years.
If spending plays out as expected into year end 2016, then construction jobs may begin to grow faster in late 2016. However, availability could have a significant impact on this needed growth.
Availability already seems to be having an effect on wages. Construction wages are up 2.6% year/year, but are up 1.2% in the last quarter, so the rate of wage growth has recently accelerated. The most recent JOLTS report shows we’ve been near and now above 200,000 job openings for months. With this latest jobs report, that could indicate labor cost will continue to rise rapidly.
As wages accelerate, also important is work scheduling capacity which is affected by the number of workers on hand to get the job done. Inability to secure sufficient workforce could impact project duration and cost and adds to risk, all inflationary. That could potentially impose a limit on spending growth. It will definitely have an upward effect on construction inflation this year. If work volume accelerates, expect labor cost inflation to rise rapidly.
I publish a lot of analysis for various construction data. I also read many other articles posted by other pundits in the industry, including MSM news sources. What I would say regarding construction data is this; an informed knowledge of construction data and how it’s used helps you understand if some article you are reading is accurate or relevant.
What I try to do here is not only report on the latest significant construction data, but also explain how the data must be used to make accurate and valuable analysis.
Here’s just three examples of how news analysts get it wrong:
> Post new construction starts as if those numbers represent construction spending.
A new start this month worth $10 billion adds a huge amount to the starts this month and will most certainly drive up the mo/mo and yr/yr starts numbers. But that new project could take 24 or 36 or 48 months to complete, so we can’t discern the impact on spending until we cash flow the value of the project which gives us the spending over its complete life span. In any given month the total amount of all spending is the summation of the spending this month from all the projects still ongoing that have started in previous months. Spending next month is 95% dependent on the flow of projects that started over the last 24 or 36 months.
> Suggest that two to three months of declines in spending indicates a downturn.
One of the biggest factors that determines spending this month is the values contributed this month from all the previous starts not yet completed. In a sector such as nonresidential buildings, in which the average duration of a project might be 24 months, the amount of spending this month gets some contribution from projects that started in each of the previous 24 months. One of the greatest influences on spending in any given month is the fluctuation (which could have occurred many months ago) in the amount of starts. So sometimes when we see a monthly spending dip it has nothing to do with a current declining trend in overall spending, but might have more to do with erratic new starts up to 18-24 months ago.
Starts can be quite erratic. Although we might see annual starts climbing at a modest rate of 6% or 7% per year, within that year we might see starts increase or decline by 50% or 100% from month to month. This is normal. But what it does to spending, particularly when a very large volume of project spending (from some month in the past that had huge new starts) finishes and drops out of the current spending, it causes dramatic fluctuations from month to month. Much of what we see in month to month changes in spending was predetermined months ago by the pattern of starts.
Normal rates of new starts, if always constant in growth rate, would create a constant rate of growth in spending. Erratic rates of change in starts create erratic changes in spending when those projects come to an end.
> Compare current $ this year to any $ from years past, without taking inflation into consideration.
I recently read an article that claimed construction was back to pre-recession levels. What really was being identified in that article was that current 2016$ were back to the level of current 2006$. That’s like saying $100 today buys you the same products that $100 bought you in 2006. I bet it wouldn’t be too hard to find a few examples where that would not be true.
Comparisons of dollars over time almost always need to be made using constant dollars, that is, adjusted for inflation and all converted to the same point in time, usually today. Sure spending today is up more than 50% off the bottom and in current dollars is higher than the previous peak in 2006. But if we adjusted those 2006$ for inflation the dollars spent in 2006 would be worth much more today. Although current dollars are now higher than any time in the past, after adjusting for inflation we are still 18% below peak spending.
September data for construction put-in-place was released today. Year-to-date (YTD) spending is up 4.4% over last year. With September first data release at $1,150 bil SAAR, this seems to establish that we’ve clearly passed a forecast dip in spending that bottomed in April and May. We’ve now had 4 months of spending up from the previous quarter and all up from the same respective months in 2015.
One thing that stands out in the data; so far every month in 2016 construction spending has been revised upwards after the first data release, by an average of +1.2%. Checking back to Jan 2014, all but once spending was revised up after the first number released.
For the 1st eight months of 2016, six of eight times the first comparison of spending showed a decline this month vs previous month. After revisions, the final values show only one month/month comparison was down.
June data which appeared quite low at first has now been revised up by +1.8% (+$21bil saar), with most of the June revision in nonresidential buildings. Most of the July revision was to residential spending. The last three months of construction spending on average have been revised up by +1.5% each.
So, even though the first print shows September down -0.4% from August, historical data would indicate we could expect September to get revised up, perhaps by 1%+ which would result in September finishing higher than June, July or August. Of course, there is always the chance it might get very little increase, and August could still get revised.
Residential spending is up 5.7% ytd and is on track to finish 2016 at $470bil, +6.6% over 2015. Last year, peak spending was in September, then residential spending dropped slightly in Q4 2015. This year I expect 2016 spending to peak in Q4, so we should see ytd performance get better as we approach year end. Cash flow from new starts indicate growth of 9% in 2017 spending.
Total Nonresidential spending is up 3.6% ytd, on track to finish 2016 at $700 bil, up 4.2% over 2015. Almost all the 2016 growth is in nonresidential buildings, not infrastructure.
For the 3rd quarter 2016, compared to the same quarter in 2015, nonresidential spending is up only 1%, but the spending patterns are not apparent unless we separate nonresidential buildings from non-building infrastructure.
For the 3rd quarter 2016, compared to a year ago, nonresidential buildings spending is up 7% and non-building infrastructure is down 6%.
Nonresidential Buildings spending is up 8% ytd through September, led by Office, Lodging and Commercial Retail markets. We should finish 2016 up 8% with a total at $410 billion vs. $379 billion in 2015. Total sector growth for the last three years is 35%. I’m predicting 2017 spending for Nonresidential Buildings will increase 8%, led by Educational and Office spending.
The market share percent of total nonresidential buildings for each market is:
educ=22%, mnfg=19%, comm=18%, offc=17%, hlthcr=10%, lodg=7% and amus/rec=5%.
Office construction spending 2016 growth will be 20%+, now greater than 20%/yr for three consecutive years. At 17% market share, by far it is the largest contributor to nonresidential buildings spending growth in 2016, contributing +3.7% growth.
Lodging is expected to finish 2016 up 26% and has averaged greater than 25%/yr growth for four years. But lodging has only 7% market share, so contributes only +1.8% growth to nonresidential buildings.
Commercial Retail is up 9% with 18% market share and so contributes +1.6% to overall nonresidential buildings growth. For the three-year period 2012 to 2014, commercial averaged 14% growth.
Educational spending, up 5% at 22% of the market, contributes +1.1% to overall nonresidential buildings growth in 2016. Educational spending should finish 2016 up 6%.
Manufacturing buildings present a unique situation in 2016. Manufacturing is down -2.4% ytd. At 19% market share, that reduces total nonresidential buildings growth by -0.5%. On the surface, manufacturing is lowering total nonresidential buildings growth. Although manufacturing spending is down, it’s still very high, so it’s impact should not be viewed as negative to the overall sector. Spending increased 50%+ in 2014 & 2015. Spending in 2016 will still be the 2nd highest year on record, down only slightly from 2015 but still more than 30% higher than 2014 and more than 50% higher than 2013.
Educational spending is 80% public and 20% private. In public markets educational is only up 4% ytd, but in private markets it’s up 10%. Private spending is driving total educational to $89 billion for 2016, up 6% from 2015. 2016 will be the best year since 2008. 2017 may reach 7% to 8% growth.
60% of all public work is infrastructure. Education accounts for 25% of public work. Educational is by far the largest building type in public work. All the remaining building types contribute only 2% to 4% each.
We are currently at what I expect could be the 2016 nonres bldgs spending peak, with very little gains across Jul-Aug-Sep-Oct. Nonres bldgs spending may flatten or drop for several months before resuming the climb. This drop may be in large part due to uneven starts from the end of 2014 and beginning of 2015, a period when starts were abnormally high, that are now finishing and dropping out of the monthly spending values. Usual normal growth patterns do not fill the void left when abnormally high volume of projects finish.
Nonbuilding Infrastructure spending is down 1% ytd. Cash flows from starts predicted this drop. The biggest negative drivers are Transportation, Sewage/Waste Disposal and Water Supply, each contributing more than 0.5% to the total decline. Power, the largest infrastructure market at 33% of total, is up 4% ytd so adds +1.33% to growth, tempering some of the declines. Spending in 2016 will reach $292 billion, down less than 1% from 2015. Growth resumes in Q1 2017.Although new starts in 2016 will finish down 10%, starts in 2015 were so high that 2016 will still be a good volume of new starts. Cash flows from all existing starts are predicting 2017 will be a record year for spending on infrastructure, up more than 6% from 2016.
My forecast for total spending in 2016 is $1,170 billion, up 5% from 2015. I expect 8% growth in 2017.
See also this post from October Starts Point to Robust 2017 Spending