Home » Productivity
Category Archives: Productivity
1-25-21 What impacts should we expect on Construction Inflation in 2021?
In April 2020, and again in June 2020, I recommended adding a minimum 1% to normal long-term construction inflation (nonres longterm inflation = 3.75%), to use 4% to 5% for 2020 nonresidential buildings construction inflation. Some analysts were suggesting we would experience deflation. Deflation is not likely. Only twice in 50 years have we experienced construction cost deflation, 2009 and 2010. That was at a time when business volume was down 33% and jobs were down 30%. In 2020, volume dropped 8% from Feb to May and we’ve gained half that back by Dec. Jobs dropped 14%, 1,000,000+ jobs, in two months! Now volume is still down 4% and jobs are down 2% from Feb peak. We’ve gained back 850,000 jobs. But also, we’ve gained back more jobs then volume. That’s inflation.
Volume drops another 5% in 2021, all nonresidential, and then another 3% in 2022. Jobs could drop overall 8%-10% for all of 2021-2022, 500,000 to 700,000 jobs.
Even though material input costs are up for 2020, nonresidential inflation in 2020 remained low, probably influenced by a reduction in margins due to the decline in new construction starts (-24%), which is a decline in new work to bid on.
Volume = spending minus inflation.
Residential business volume dropped 12% from the January 2020 peak to the May bottom, but has since recovered 22% and now stands at a post Great Recession high, 10% above one year ago. Although residential spending remains near this high level for the next year, volume after inflation begins to drop by midyear. For the year 2020, Residential Building Materials Inputs are up 6.2%. See PPI charts. Sharply higher lumber prices have added more than $17,000 to the price of an average new single-family home since mid-April ($24,000 as of 3-30-21). Residential inflation averaged 5.1% for 2020. (UPDATE 3-30-21 – Single Family home prices increased 11% since March 2020. Lumber cost is now 3x what it was in March 2020. These will both impact cost to build SFH).
The U.S. Census Single-Family house Construction Index is up 6% from Nov 2019 to Nov 2020. The index increased 4% in the last 5 months. The index is increasing, now thru February 2021 is up 6.7% for the last 12 months. https://www.census.gov/construction/nrs/pdf/price_uc.pdf
Nonresidential volume has been slowly declining and is now down 10% from one year ago. By 3rd quarter 2021, nonresidential buildings volume is forecast down another 15% lower than December 2020, or 25% below the Feb 2020 peak. This tracks right in line with the 24% decline in new construction starts in 2020. Most of the spending from those lost starts would have taken place in 2021, now showing up as a major decline in spending and work volume. Nonresidential inflation for 2020 dropped to 2.5%, the first time in 7 years below 4%. It’s expected to increase in 2021.
The Producer Price Index tables published by AGC for year-end 2020 https://www.agc.org/sites/default/files/PPI%20Tables%20202012.pdf shows input costs to nonresidential buildings up about 3.5% to 4.5% for 2020, but final costs of contractors and buildings up only 1% to 2%. This could be an indication that, although input costs are up, final costs are depressed due to lower margins, a result of fewer projects to bid on creating a tighter new work available environment which generally leads to a more competitive bidding environment. This could reverse in 2021 as the volume of work to bid on in most markets begins to increase.
As of March 2021, PPI for materials inputs to construction is up 12% to 14% yoy, measured to last March before the bottom dropped out. The Building Cost Index for final cost to owner is up only 2%. Construction inflation is very different right now for subcontractors vs general contractor/CM. https://www.agc.org/learn/construction-data
The Turner Construction Cost Index (nonresidential buildings) for Q1-Q2-Q3 is +1%, -1%, -0.5%, effectively reporting the index down -0.5% year-to-date. But the Turner index year-to-date average (avg Q1+Q2+Q3=1179) is still 2.6% higher than the average of Q1+Q2+Q3 2019 and 2% higher than the avg for all of 2019 (1156). So, while the index appears to show no gains in 2020, through the first nine months of 2020 it is up 2.6% above the average of the same months in the 2019 index. http://turnerconstruction.com/cost-index
The Rider Levitt Bucknall nonresidential buildings index average index for 2020 (through October 1, 2020), although up less than 1% in the last 6 months, is up 3.5% from the average 2019 index. https://s28259.pcdn.co/wp-content/uploads/2020/07/Q2-2020-QCR.pdf
R.S.Means quarterly cost index of some materials for the 4th quarter 2020 compared to Q1: Ready-Mix Concrete -1.8%, Brick +10%, Steel Items -1% to -5%, Framing Lumber +32%, Plywood +8%, Roof Membrane +5%, Insulating Glass +12%, Drywall +3%, Metal Studs +23%, Plumbing Pipe and Fixtures +1%, Sheet Metal +20%. https://www.rsmeans.com/landing-pages/2020-rsmeans-cost-index
U.S. manufacturing output posts largest drop since 1946. Think of all the manufactured products that go into construction of a new building: Cement, steel, doors, frames, windows, roofing, siding, wallboard, lighting, heating systems, wire, plumbing fixtures, pipe, valves, cabinets, appliances, etc. We have yet to see if any of these will be in short supply leading to delays in completing new or restarted work.
There have been reports that scrap steel shortages may result in a steel cost increase. Scrap steel prices are up 27% in the last quarter and up 40% for the year 2020. Scrap is the #1 ingredient for new structural steel. The U.S. steel industry experienced the most severe downturn since 2008, as steelmakers cut back production to match a sharp collapse in demand and shed workers. Capacity Utilization dropped from 82% in January 2020 to 56% in April. In mid-August, CapU was up to 61%, still very low. As of January 23, 2021 CapU is up to 76%, well above April’s 56% but still below desired levels. Steel manufacturing output is still down compared to pre-covid levels. Until production ramps back up to previous levels there may be shortages or longer lead times for delivery of steel products.
Steel Prices at mill in the U.S. are up 60% to 100% in the last 6 months. All prices are 50% to 75% higher than Feb 2020. http://steelbenchmarker.com/files/history.pdf . This is mill price of steel which is about 25% of the price of steel installed. What affect might a steel cost increase have on a building project? It will affect the cost of structural shapes, steel joists, reinforcing steel, metal deck, stairs and rails, metal panels, metal ceilings, wall studs, door frames, canopies, steel duct, steel pipe and conduit, pumps, cabinets and furniture, and I’m sure more. Assuming a typical structural steel building with some metal panel exterior, steel pan stairs, metal deck floors, steel doors and frames and steel studs in walls, then all steel material installed represents about 14% to 16% of total building cost. Structural Steel only, installed, is about 9% to 10% of total building cost, but applies to only 60% market share being steel buildings. The other 6% of total steel cost applies to all buildings. https://www.thefabricator.com/thefabricator/blog/metalsmaterials/steel-prices-reach-levels-not-seen-since-2008 At these prices, if fully passed down to the owner, this adds about 1.5%-2% to building cost inflation. With demand in decline for nonresidential buildings, I would expect to see all these steel price increases recede. Also, take note, as of January 2021, none of this steel price movement appears captured in the PPI data or RSMeans data.
Contractors have been saying they have difficulty acquiring the skilled labor they need. This has led to increased labor cost to secure needed skills. I expect the decline in nonresidential work volume in 2021 to result in as much as a decline of 250,000 nonresidential jobs in 2021. This results in labor available to fill other positions.
This SMACNA report quantifies that labor productivity has decreased 18% to meet COVID-19 protocols. https://www.constructiondive.com/news/study-finds-covid-19-protocols-led-to-a-7-loss-on-construction-projects/583143/ Labor is about 35% of project cost. Therefore, just this productivity loss would equate to -18% x 35% = 6.3% inflation. Even if, for all trades, the average lost time due to COVID-19 protocols is only half that, the added inflationary cost to projects is 3% above normal. But that may not remain constant over the entire duration of the project, so the net effect on project cost would be less.
Post Great Recession, 2011-2020, average nonresidential buildings inflation is 3.7%. In 2020 it dropped to 2.5%, but for the six years 2014-2019 it averaged 4.4%. Residential cost inflation for 2020 reached 5.1%. It has averaged over 5% for the last 8 years. The 30-year average inflation rate for nonresidential buildings is 3.75% and for residential it’s over 4%.
This survey of members by AGC https://www.agc.org/sites/default/files/2021_Outlook_National_1221_.pdf just published provides some insight into construction firms outlook for 2021.
Almost every construction market has a weaker spending outlook in 2021 than in 2020, because approximately 50% of spending in 2021 is generated from 2020 starts, and 2020 nonresidential starts are down 10% to 25%, several markets down 40%. Nonbuilding starts are down 15%, but will increase 10% in 2021.
Typically, when work volume decreases, the bidding environment gets more competitive. We can always expect some margin decline when there are fewer nonresidential projects to bid on, which typically results in sharper pencils. However, if materials shortages develop or productivity declines, that could cause inflation to increase. We can also expect cost increases due to material prices, labor cost, lost productivity, project time extensions or potential overtime to meet a fixed end-date.
Constant $ = Spending minus inflation = Volume
Many projects under construction had been halted for some period of time and many experienced at least short-term disruption. The delays may add either several weeks to perhaps a month or two to the overall schedule, in which case, not only does labor cost go up but also management cost goes up, or it could add overtime costs to meet a fixed end-date. Some of these project costs have yet to occur as most would be expected to add onto the end of the project.
Some projects that were put on hold (nonresidential buildings starts in 2020 dropped 24%) just prior to bidding in 2020 may now re-enter the bidding environment. The rate at which these projects come back on-line could impact the bidding environment. If several months worth of projects that delayed bidding last year all come onto the market at once, or at least all in a more compressed time span than they would have, the market could be flooded with work and bidding contractors now have more choice, can bid more projects than normal and could potentially raise margins in some bids. This would have an inflationary effect. Also, there can be difficulty in starting many projects at the same time, rather than more staggered starts. It burdens subcontractors and suppliers with too much of the same type of work all going on at the same time. This could exacerbate labor issues and could lead to project time extensions.
The hidden inflationary costs of bidding environment, project time extensions, potential overtime and lost productivity haven’t all yet appeared in the data. Some of these could still add to 2020 inflation. Also, the huge loss of new starts in 2020, which meant fewer projects to bid on in 2020, probably reduced margins in 2020. Nonresidential starts are projected to increase 4% in 2021, so that could lead to some recovery of margins, however, even with 4% growth in new starts, that comes after a 24% drop in 2020, so remains still 20% below 2019. Total volume of work is declining and new projects available out to bid is still depressed, so pressure on margins still exists.
update 4-15-21 Although materials cost inflation will be higher, I expect non-residential buildings inflation final cost in 2021 to range between 3.5% to 4.0%, with potential to be held lower. Subcontractor costs, such as for steel or lumber, could range much higher due to huge material cost increases. All the downward pressure on nonresidential inflation is on margins. There is currently 20% less work to bid on than in Q1 2020. Expect 2021 residential inflation of 6% to 8% (updated 3-30-21) for residential work with potential to push slightly higher.
See Construction Inflation Index Tables for indices related to Nonbuilding Infrastructure work.
The tables below, from 2011 to 2020 and from 2015 thru 2023, updates 2020 data and provides 2021-2023 forecast. The three sectors, highlighted, are plotted above. (4-16-21 The tables do include updated residential costs and updated nonresidential inputs).
NOTE, these tables are based on 2019=100. Nonresidential inflation, after hitting 5% in both 2018 and 2019, and after holding above 4% for the six years 2014-2019, is forecast to increase only 2.5% in 2020, but then to 3.8% in 2021 and hold near that level the next few years. Forecast residential inflation for the next three years is level at 3.8%. It was only 3.6% for 2019 but averaged 5.5%/yr since 2013 and returned to 5.1% in 2020 Recent materials costs and volume of work could impact residential inflation and push it higher.
How to use an index: Indexes are used to adjust costs over time for the affects of inflation. To move cost from some point in time to some other point in time, divide Index for year you want to move to by Index for year you want to move cost from. Example : What is cost inflation for a building with a midpoint in 2022, for a similar nonresidential building whose midpoint of construction was 2016? Divide Index for 2022 by index for 2016 = 110.4/87.0 = 1.27. Cost of building with midpoint in 2016 x 1.27 = cost of same building with midpoint in 2022. Costs should be moved from/to midpoint of construction. Indices posted here are at middle of year and can be interpolated between to get any other point in time.
All forward forecast values, whenever not available, are estimated by Construction Analytics.
12-6-19 plots updated to include Nov jobs and Oct spending.
Construction Spending IS NOT Construction Volume.
I read an analyst report this week that stated construction jobs growth isn’t keeping pace with construction volume growth. The reference appeared to be to construction spending. That fails to apply inflation to convert construction spending to construction volume, so compares apples to oranges. Spending must be adjusted for inflation to get real volume growth. Jobs MUST be compared to volume.
For over two years now, construction volume growth has not supported construction jobs growth we’ve seen. I expected jobs growth to slow down. I’ve been saying this for over a year. This sure looks like it.
For 2018 jobs growth averaged over 300k. Since January 2019 the rate of jobs growth has dropped from 300k to 150k.
Current projected new starts data IS NOT supporting construction volume growth for the next 2 yrs. Growth of 3%/yr in non-building infrastructure will be offset by declines in residential buildings and flat nonresidential buildings. Therefore, there is no real volume support for jobs growth.
This plot adjusts construction spending by taking out inflation to get real construction volume growth. Last year of real volume growth was 2016. Yet jobs continue to climb. This can’t continue. The plot above shows it has slowed.
Construction jobs growth has slowed considerably over last 2Q, as expected. While construction jobs are up about +150k in last year, jobs (through Nov) increased only +48k in the last 7 months. I’m expecting this trend to continue. In fact, I wouldn’t be the least bit surprised to see in the near future some months when construction jobs decline. The fact is, construction volume simply does not support jobs growth.
Total construction volume, spending after accounting for inflation, has been down for 5 of the last 6 quarters. Volume peaked from Q1 2017 to Q1 2018, but the last year of real volume growth was 2016. Volume is flat or down while jobs continue to rise. This can only mean contractors will be at risk of being top-heavy jobs if a downturn comes.
Caution is advised if putting emphasis on construction JOLTS, which has been climbing to new highs. From mid-2006 to mid-2007, JOLTS reached near the then all-time high. But construction volume, starting in mid-2006, was already on the downward slope. Volume peaked in early 2006 and fell 10% by mid-2007. Construction did not begin shedding jobs until late 2006, but mid-2007, job losses were well underway. Within 12 months, more than 500,000 jobs were gone. Within 18 months, construction jobs were down 1.5 million.
Construction spending annual rate will increase by 3% in the next 12 months, but volume in constant $ after inflation will remain flat. In Q42020-Q12021 spending slows to less than inflation, so volume begins a modest decline. Growth of 3%/yr in non-building infrastructure will be offset by declines in residential buildings and flat nonresidential buildings. Jobs will continue to grow and spread the imbalance even more.
The construction jobs slow down has been in the cards for a long time. With all the talk of skilled labor shortages, there’s been little discussion of the unsustainable excess jobs growth. Maybe it’s about time to change the conversation.
In the 24 months from May 2016 to May 2018, Construction Volume went up 3.0%. Jobs went UP by 8%, 500,000 jobs. Spending in that 24 month span increased by just over 12%, but inflation for that period across all construction averaged 9%, hence real volume increased only 3%. That’s a $35 billion increase in volume, enough new work to support 175,000 to 210,000 new construction jobs.
JOLTS (Job Openings and Labor Turnover Survey) job openings went up from 2.4% to 3.0%, up 50,000 openings. Jobs growth exceeded volume growth by more than double and yet job openings went up!
Not only did jobs growth of near 8% far exceed that needed to support the growth in new work, but also, because jobs growth was so strong, it should have reduced job openings.
What’s wrong with this picture?
Pretty obvious the numbers just don’t add up. First, since construction spending is always later revised up, in recent years by 2%, let’s be generous and assume spending will get revised up by 2%, and let’s keep inflation the same. That would result in a 5% increase in volume or closer to $60 billion in volume. That would support 300,000 to 360,000 new jobs, a need still well below the actual growth in jobs of 500,000.
No matter how we look at it, even generously supposing spending will later increase by 2%, jobs have increased greater than volume of work.
Companies predict job openings based on positions they need to fill within 30 days. But, what if their judgement of positions they need to fill is determined based on what they anticipate from increases in revenue, without taking inflation into consideration. Since revenue also includes inflation, which adds nothing to business volume, that would overestimate the need for new jobs. We’ve seen this before, in the last expansion.
2003-2006 construction spending increased by 35%, the most rapid increase in spending in over 30 years. But construction inflation during that four year period totaled over 30%, the most for four consecutive years dating back to 1978-1981. After adjusting for inflation real volume in 2003-2006 was up by less than 5%. Considering how high spending was and how much it felt like growth, there was surprisingly little. That did not hold back jobs expansion.
Construction firms added 15% to jobs, or 1,000,000 jobs during this period, more than 3x the actual need. Job Openings in the JOLTS report increased 100%+, from 100,000 to over 200,000. Firms hired far more than needed and kept increasing the report of job openings, even though they had already hired far more than required. In 2006, housing starts dropped 15%, residential spending dropped 25%, but residential jobs still increased by 6%. From 2003 to 2006, spending on nonresidential buildings increased by 20%, all of it inflation. Volume remained stagnant these four years, however jobs increased by 10%.
Clearly the increases in jobs during this period correlate more with spending than real inflation adjusted volume growth. This four-year period registered the largest productivity decline in over 30 years because the rate of jobs growth was much faster than volume growth.
For 2018-2019-2020, construction spending is currently forecast to increase 6.7%, 3.0% and 4.2%. But after adjusting for inflation, real construction volume is predicted to increase only in 2018 by about 2%. For 2019-2020 volume declines or remains flat.
An argument could be made that JOLTS openings is dependent on firms outlook for growth in the near future. For that, let’s look at predicted volume growth in 2nd half 2018 and in 1st half 2019. It is predicted spending will increase 1.5% in the 2nd half vs 1st half 2018. But adjusted for inflation, volume will decline by 1%. Likewise, for the 1st half 2019, although spending will increase, inflation will outpace spending and real volume will decline 1%. There is nothing in past data or forecast that would support an increase in forecast job openings.
See also What Jobs Shortage? 7-6-18 for related info.
Could it be that some firms are anticipating job needs based on spending, not on volume? Could it be that these firms are not adjusting revenues for inflation to get volume before using the data to prepare a business plan? This is not entirely anecdotal. In several presentations I’ve given over the years I’ve asked the audience, How many of you plan your business needs on your revenue? In a show of hands at a presentation to NHAGC, a large portion of the audience raised their hand.
If your construction company revenues are up 6% in a year when inflation is 5%, then your net volume is up only 1%. Your company jobs growth required is only 1%.
You cannot ignore the impact of inflation when forecasting jobs need.
Construction Overtime – A Common Miscalculation
You never get full production out of all overtime hours worked. A common miscalculation when applying overtime overlooks productivity losses.
Let’s say we have a project that has 100 manweeks of productive work (100mw x 40hrs = 4000 manhours) remaining on the schedule to completion, but that we absolutely must finish the job is less time. Also, let’s say we modify the work week from 5 days 8 hours = 40 hours/wk to Overtime (OT) 6 days 10 hours = 60 hours/wk. A simple calculation indicates that if we add 50% more hours per week (60hrs vs 40hrs), we could finish the job in 1/3 less time.
- Original plan = 4000 manhours / 40 hrs/week/man = 100 manweeks
- Revised plan = 4000 manhours / 60 hrs/week/man = 67 OT manweeks
- Time saved = (100–67)/100 =33/100= 33% time saved, 33 mwks saved
- Cost added would be +20%. See example of cost calculation below.
But, unfortunately, that would not be correct. That would have to assume no OT productivity losses. You won’t get 60 productive hours out of a man in a 6-10s 60-hour OT workweek. You will get only 50 productive hours.
Productivity loss graphic from Applied Cost Engineering, Clark and Lorenzoni, Marcel Dekker, Inc., 1985.
Yes, you still pay for all hours and the man is still on the job for 60 hours, but work progress slows as workers are kept on the job for longer periods. So how much time would be saved on the schedule?
Revised plan productivity 4000 manhours work / 50 productive hrs per week per man = 80 OT manweeks to completion.
Time saved = (100 – 80) / 100 = 20/100 = 20% time reduction or 20 mwks saved, not 33.
What did we get from this application of overtime compared to the original?
- 20 mnwks LESS of normal 40hrs =20×40= 800hrs less at normal 1x rate
- 80 mnwks at 20hrs/wk at OT, 1.5x rate =80×20= 1600hrs more at 1.5x rate
- Net cost 1600 x 1.5 – 800 x 1 = 1600 equivalent extra cost hrs over base 4000.
- Time saved (100-80)/100 = 20%
- Cost increased 1600/4000 = 40%
This simple example shows the full hourly time savings is not realized due to lost productivity plus many of the hours worked are at a higher cost. Though the initial basic OT estimate forecast 33% time saved at 20% extra cost, that scenario actually saved only 20% time and added 40% cost, double the initial budget.
If this was initially a 30 month project, with approximately 35% of the cost in labor, then overtime saved 6 months time, but added 15% inflation to the total cost.
There’s a significant difference in the original un-adjusted OT estimate of time/cost versus the OT time/cost analysis for nonproductive hours. That would be a serious mistake in estimating and could have serious cost implications against the budget.
This will vary with the OT scenario selected or any other data set used, but generally the more days and longer hours worked, the higher the extra cost ratio. Of course, a better way to accomplish a tightened schedule might be to add a second shift rather than work men longer hours. However, in times of restricted labor supply that might not be feasible.
See this blog post for OT productivity loss rates Overtime Isn’t Always What It Seems – Lost Productivity Construction
The last time construction jobs and workload were balanced was 2005. From 2006 through early 2011, workload dropped 15% greater than the decline in jobs. In other words, compared to 2005, contractors started the post-recession period in 2011 with 15% less workload on hand compared to the number of workers kept on staff and that resulted in the period 2006-2011 posting the largest productivity decline ever recorded.
For a discussion on data plotted 2001 to 2011, see this post Jobs vs Construction Volume – Imbalances. In the 2001-2011 plot above, jobs and workload are set to zero baseline in Jan 2001. This shows all of 2001 through 2004 that jobs/workload was balanced. The gap between the red and the blue lines above is the variance from zero change in Jobs/Workload balance. By Jan 2011 there was a 15% workload deficit.
The 1st quarter of 2011 was a dramatic turning point. Both jobs and work volume began to increase. To visualize the variance since Jan 2011, the following plot resets jobs and workload to zero baseline in Jan 2011.
From Jan 2011 to Jun 2015, construction volume increased 24% in 4 1/2 years. Staffing output increased 19% in the same period. Contractors may still feel the effects from not being able to grow staff at that same pace as volume during that period. However, we did see the larger work volume increases make up 5% of the 15% workload deficit from the previous period 2006-2011, but it loses sight of the fact that after almost five years we had not recouped the entire lost work output from all the other 10% staff imbalance that still remained.
Work output is defined as jobs x hours worked. Construction volume is defined as spending minus inflation.
From Jul 2015 to Oct 2017, volume increased just over 1% but jobs output grew by almost 7%. During that two year period, new jobs created plus the change in hours worked by the entire workforce grew 6% more than workload. Jobs increased greater than construction volume increased. The plot shows most of that variance occurred in 2015.
Shifting the time periods slightly gives another impression of the data, overall not much different. In discussions about Construction skilled labor shortages, it’s important to understand, both construction spending and volume are at record growth levels and jobs, since recession, and in last 3 yrs, have matched volume growth.
Overall, in the seven-year post-recession period Jan 2011 to Oct 2017, volume increased 25% and jobs output increased 26%. There seems very little room to be calling this a jobs shortage. Of course, this does not address skills.
So here we are most of the way through 2017 and if we look back at the last 11 years, not only are jobs once again increasing faster than workload, but also in total since 2005 we still have 14% staff that would need to be absorbed by new workload to return to the previous jobs/workload productivity balance.
Maybe it’s time we stop calling this a jobs shortage and start referring to it as a productivity challenge that needs to be turned around.
For an expansion of more information on this topic see Jobs vs Construction Volume – Imbalances posted 8-8-17. Included is the 2001-2011 plot that explains all of 2001 through 2011.
Also, Feb 208 article breaking out residential and nonresidential sectors shows surplus in nonres and deficit in residential Residential Construction Jobs Shortages
Attached PDF of my Forecasting presentation delivered 5-22-17 at Advancing Building Estimation in Houston
A few bullets from this presentation
- Construction Starts is not construction spending
- Cash flow = Spending = Revenue
- Revenue is not Volume of work
- Spending minus inflation = Volume
- Understand what’s in an Index to avoid misguided inflation adjustments
- We can’t ignore productivity
- Spending activity has just as much influence on inflation as labor and material cost.
Slides in this presentation come from the following articles:
Jobs growth slowed in the last two months adding only 6,000 construction jobs since February. However, a longer term look at jobs x hours worked vs volume growth gives better information.
In the following plot Jobs (red line) = # of jobs x hours worked and Construction Volume (blue line) = construction spending in constant $ (adjusted for inflation). Unless we make these two adjustments we cannot compare jobs to construction spending and get any meaningful analysis from the data.
I’ve written about this in-depth in these two articles.
You can see in the plot above from Jan 2011 to Mar 2013 both jobs growth and volume growth balanced. Then again by August 2014 jobs growth caught up to volume growth. It was the period from Aug 2014 to Jul 2015 when volume took off and climbed much faster than jobs growth. But then, since July 2015, jobs have been increasing faster than construction volume growth.
In a plot of this information back to 2005, it would show that by the end of 2010 there were already excess jobs. That is discussed in the attached articles. During the expansion, firms hired more employees than real work volume could support, then during the recession, firms held onto far more staff than was required to perform the available declining work volumes. So the chart above would start 2011 with an excess of jobs and really we needed to see work volume increase faster than jobs starting in 2011.
Long term, having started 2011 with not enough volume to support the remaining staff, we see two periods of growth in which jobs and volume were balanced, only one period where volume exceeded jobs growth and then this latest period, for the last 21 months, in which jobs are growing faster than volume.
There are many reports of job shortages and they appear to be genuinely accurate assessments, primarily regarding some very specific skilled labor positions. However, long term jobs vs volume data shows there is far more in play than not enough workers to hire. In fact, for the last 21 months, hiring has exceeded workload and that simply does not indicate an overall worker shortage.
I’ve been saying for a long time the data doesn’t show a construction jobs shortage.
In total, construction jobs have been increasing faster than construction volume (spending minus inflation). But, to get a better picture we need to look at jobs vs volume by sector, Residential and Nonresidential. Then we need to look at history.
Since 2009, RESIDENTIAL volume has increased 49%, jobs increased 22%. This is partly explained by absorption of excess staff retained during recession.
From 2006 to 2009 volume decreased 53% but jobs decreased only 36%, leaving a significant amount of excess jobs.
It looks like from 2009 to 2016 there has not been enough jobs growth to support the volume growth, BUT…
Residential net changes just since 2006, volume is down 29% while jobs are down 22%. We are not nearly back to pre-recession productivity.
Since 2009, NONRESIDENTIAL BUILDINGS volume is down by 10% but jobs are up 13%. By no means, if we look at just these 7 years, does this look like a jobs shortage.
Even previous years imbalance would not account for a need to add that many jobs. From 2006 to 2009 volume increased 2% but jobs decreased 15%. In a previous report Is There a Construction Jobs Shortage? I explained why this may occur following a prior top-heavy jobs expansion during a period of high inflation.
Nonresidential net changes just since 2006, volume is down 8% but jobs are down only 3%. Again, we are not nearly back to pre-recession productivity.
For both residential and nonresidential buildings, comparing post-recession growth to pre-recession 1996-2006 $ Put-In-Place per Job, productivity is down 21%, or we currently have 100/(100-21) = 27% more jobs now than it took before to get the same amount of work done.
If the current construction expansion period is viewed as having a jobs shortage, that claim demands that we must accept, since pre-recession, productivity has declined by 21% and the reason there is now a jobs shortage is that it takes 27% more jobs to put in place construction than it did on average from 1996 to 2006.
In my opinion, that’s a harder pill to swallow than a jobs shortage.
For more related to this discussion see Is There a Construction Jobs Shortage?
The imbalance between construction spending and construction jobs is nothing new. It’s been going on for years. It reflects more than just worker shortages. It captures changes in productivity due to activity. It also helps explain why sometimes new jobs growth rates do not follow directly in step with spending growth. A big part is that it reflects hiring practices. That imbalance can be affected by either over or under-staffing and that can be affected by inflation.
2000-2008 The Expansion
For the 1st several years, nonresidential construction spending was flat or down. Then for two years spending was up only slightly, but constant $ volume (spending inflation adjusted) had actually decreased. Nonresidential jobs fell from 2001-2003 but then grew for several years during this period when constant $ volume was decreasing, creating productivity losses.
On the other hand, residential spending grew 80%, but after adjusted for inflation, volume grew only 23%. Most staffing increases during this period were for residential construction and jobs/volume growth was pretty consistent. Residential saw mixed productivity during this period. In 2006 residential volume had already started declining.
It is not uncommon when work is plentiful that productivity declines. In 2004-2005, spending increased by 24%, but inflation was hovering around 8% to 9%/year. Constant $ volume (spending after inflation) increased by only 6%. Jobs grew faster, by 9%. Net productivity decline.
In 2006, nonresidential work was starting to take off, increasing 45% from 2006 though 2008. During this period jobs increased by only 8% and volume added 16%. Excess volume was able to absorb a good portion of the jobs/volume imbalance from 2000-2006.
See the line chart below “Productivity = Annual $ Put-In-Place per worker. These up or down periods for each of these sectors discussed here can easily be seen in rising or falling $PIP volume on that chart, sectors plotted separately. The bar graph “Total Annual Productivity Change”, is the composite total of the three sector graphs.
Net volume in 2006 declined, but jobs increased another 5%. For the three-year period 2004-2006, spending increased by 28%, but after inflation, real volume increased by less than 5%. Jobs increased by 14%. Productivity declined by nearly 10%.
Heading into 2007, residential firms had excess staff, as measured by the negative imbalance of jobs/volume. Compounding productivity issues, when spending started to decrease significantly, it took longer for companies to downsize their workforce. The workforce was not reduced to match the volume of work lost. Residential construction was first to show the strain, already having started to decline in 2006 and continuing to decline through 2009.
2006-2010 The Residential Recession
Residential construction was 1st hit by the recession in early 2006. For the 4 years 2006-2009, residential volume dropped 55%. It remained flat for two more years, down a few more percent. Over six years starting 2006, residential jobs dropped only 40%.
The Annual $ PIP line chart above shows that for 2006-2009 there were only residential losses, or negative balance between jobs/volume. Both nonresidential sectors were improving slightly at the time. The total negative bars in those years is entirely due to residential.
2009-2011 The Nonresidential Recession
Nonresidential Buildings construction didn’t fall into recession until 2009. In the two years 2009-2010, nonresidential buildings lost over 30% in volume but only 22% in jobs.
This chart simply shows the imbalance between the number of jobs and the dollar volume of work put in place for each year compared to the year before. In a simple form that can be referred to as a change in productivity. In all these charts, jobs/year are adjusted for hours worked and dollars are always constant $ inflation adjust to 2016$.
In 2009 my chart shows a huge productivity gain. It is almost entirely due to Non-building infrastructure, which never did fall into deep recession. Combined residential and nonresidential buildings only in 2009 would have shown a net 1% gain.
2011-2012 Early Recovery
Starting 2011, firms had lost significant revenue but had retained more staff than needed. There was so much excess staff (in relation to how much total revenue was available) that almost no reasonable gains in spending could wipe away the losses in productivity. Volume improved by 1%, but hiring resumed and jobs grew by 1%. Due to excess staff still on payrolls, productivity showed a 6% decline.
For the next few years, when we look at jobs growth vs. volume growth, there is reason to believe that slow jobs growth (2011 through 2013) may not be all due to labor shortages. Although we lost more than 2 million jobs, there remained excess jobs when compared to the amount of volume that was available.
At least part of the blame for slower new jobs growth was that excess staff already on hand were being absorb by the new spending gains. For a period there was insufficient volume out in the market to support all the staff that had remained on board. Finally, there was increased revenues which would first reabsorb part of the excess labor before rehiring started.
2012-2016 The Construction Boom
It took three more years to see a significant move towards balancing jobs and real volume. In 2014, jobs increased 6% and constant volume increased 7%. For the first nine months of 2015, jobs increased 3% and volume increased 8%. This was a good productivity balance period.In the three years 2012-2014, volume increased 16% but new jobs grew by only 11%. The increased work volume absorbed a good portion of the excess staffing.
What reasons could cause contractors to think they need more staff?
One reason may be that contractors don’t typically track revenues in constant dollars, they track in current dollars. So any comparison to a previous year is to inflated data. To achieve business plan growth of 6%/year, is it necessary to grow staff by 6%/year? Not if during that period inflation is 4%. Then real volume growth is only 2%/year and new staffing needs are far less than anticipated.
Basing staffing needs on current $ revenue growth can lead to the same kind of over-staffing we saw going into the recession. In the three years 2004-2006, construction spending increased 30%. Jobs increased 16%. However, during that three-year period construction inflation was the highest ever recorded, composite inflation averaging greater than 8%/year. After inflation, real construction volume increased only 4% during that period. Hiring far exceeded the rate of real volume growth. There is the potential that contractor’s hiring could be swayed by highly inflated spending when actually volume is not as strong as thought.
From the Jan 2011 bottom of the construction recession through Dec 2016, both work output (jobs x hours worked) and volume (spending after adjustment for inflation) increased equally by 29%.
(note: BLS revisions to hours worked, issued in the 3-10-17 release changed total growth output from 29% to 30%).
There are always unequal up and down years, but this longer term period shows balanced growth returned after a tumultuous period. We were so far down on the scale after the recession it seems reasonable that we experienced this re-balancing.
Both 2014 and 2015 show productivity gains. That is unusual in that there have not been two consecutive years of productivity gains in 23 years (while my jobs data goes back to 1970, spending data goes back only to 1993).
The trend changed in October of 2015. Now when we look at jobs growth vs. volume growth, there is reason to believe that any jobs growth slow-down may be at least in part due to recent over-hiring.
2014-2016 Record Jobs Growth
In the last three years, we’ve added 840,000 construction jobs. We’ve also increased hours worked to an equivalent to 880,000 jobs, growth of 15%. That’s a faster rate of growth in three years than the 2004-2006 construction boom. To help explain that growth, real volume in 2014-2016 was far greater than the volume in 2004-2006, or any other three-year period for that matter. The last time we’ve seen jobs growth like this was 1995-1999.
2012 through 2016 is the greatest construction boom on record, whether measuring unadjusted current $ spending or constant $ real volume after inflation, flying past the 2000-2005 boom and narrowly beating out 1995-2000. And we started 2017 with backlog at a record level, so the boom continues.
5-Year Construction Booms Compared to 2012-2016
- 2012 – 2016 current $ +$377 bil +48% — constant $ +265 bil +29%
- 2001 – 2005 current $ +$314 bil +39% — constant $ +30 bil <3%
- 1996 – 2000 current $ +$254 bil +46% — constant $ +235 bil +21%
Notice how little growth actually occurred in the five-year period 2001 through 2005. While there was significant spending growth, most of it was inflation, and 90% of it was residential. During that period composite inflation increased more than 35%. Also, nonresidential construction was having a setback, dropping 15% in volume in that five years. The real story out of the 2001-2005 boom period is to compare residential work.
- 2012 – 2016 Rsdn current $ +$211 bil +83% — constant $ +146 bil +46%
- 2001 – 2005 Rsdn current $ +$280 bil +80% — constant $ +132 bil +23%
Residential inflation 2001-2005 was a whopping 47%. But, total residential spending was up 80%. After adjusting for inflation, residential still added 23% to volume during that period. During both periods, residential volume grew more than jobs, so both periods had a net productivity gain.
Also in 2001-2005, nonresidential added 3% more jobs in a five year period in which volume dropped 15%. The very high levels of inflation help explain why staff may have grown to such excess during that period. Contractors were seeing revenues grow by 20%-30% and were slowly adding jobs in a period when real volume was dropping 3% per year.With the exception of residential growth, there was a downturn in other work. New jobs increased by only 11%, but due to rampant inflation, real volume increased by less than 3%. Nonresidential contributed all the negative productivity in 2001-2005.
2014-2015 Construction Spending for the Record Books
- 2014 to 2015 current $ +$206 bil +23% — constant $ +158 bil +16%
No two consecutive years of construction come close to equaling the real volume put-in-place during 2014-2015. The two years 2004-2005 had greater growth in spending, but most of that was inflation, so had little growth in volume. In fact, we would need to consider three consecutive years to come close to 2014-2015 and the three years that comes closest is 1996-1998 and that would still be a few percent short. This volume growth is driving huge jobs growth.
From October 2015 through March 2016, jobs growth was exceptional. During that 6 month period we added 215,000 construction jobs, the fastest jobs growth period in a decade. That period topped off the fastest two years of jobs growth in 10 years. Record increases in jobs growth are not what we might expect if there is a labor shortage.
And yet, the Jobs Opening and Labor Turnover Survey (JOLTS) is the highest it’s been in many years and that is a signal of difficulty in filling open positions. But, one of the known factors during a high level of market activity (lot’s of construction work – we are at record levels) is that workers know there is another and sometimes better job just down the road. During high levels of activity, unless the current employer is paying some kind of premium to keep them, workers may leave for greener pastures. That creates a high level of job churn.
Hiring Changes Lag Volume Changes
It is important to take note that it appears the two most recent six-month surges in jobs lag the period of greatest volume growth. I noted earlier that contractor staffing changes seem to lag movements in volume.
Since Sep 2015, jobs have been increasing more than real construction volume. For much of 2014 and 2015 construction spending real volume growth was exceeding jobs growth. Spending in 2016 slowed from the all-time record levels. That’s not totally unexpected as it would be highly unusual for that record level of growth to continue. But hiring continues.
Since Sept 2015, construction volume growth (spending minus inflation) slowed or stalled and completely contrary to what one would expect in a labor shortage, new jobs growth has been exceeding volume.
- From Sep15 to Mar16 jobs increased+3.3%, volume increased +1.6%
- From Mar16 to Aug16 jobs had no change, volume decreased -3.3%
- From Aug16 to Feb17 jobs increased +2.6%, volume increased +0.10%
This most recent six-month period posted 177,000 jobs, the 3rd best for any consecutive six months since 2005-2006. Although we experienced a slow down in new jobs through the middle of 2016, that was bracketed by two of the three strongest six month growth periods in more than 10 years. For 18 months Sep’15 to Feb’17, jobs are up 7% higher than volume. For 2012-2014 volume grew 6% more than jobs.
For 2017, several economists are predicting total construction spending will increase by just over 6% (including my estimate of 6.5%). However, I’m also predicting that combined construction inflation for all sectors will increase by about 4.5%. That leaves us with a net real volume growth of only 2.0%. Therefore, for 2017, I do not expect jobs to increase by more than 2.0%, or 140,000. That number seems hard to swallow given we are already at 98,000 in the first two months. But remember, jobs have been growing faster than volume for the last 17 months. We could be due for another no-jobs-growth absorption period.
If jobs increase more than 140,000 and both spending and inflation hold to my predictions, then jobs will continue to outpace volume and that will show up on my plot as a productivity loss for 2017. Jobs have been getting ahead of volume for 17 months. Contractors may still be hiring, lagging the movement in real volume growth. It will take the next few months to see if that is the case but I would expect jobs growth to slow or stop for the next few months and I would not attribute that to labor shortages. As we’ve seen before, we should expect jobs/volume to come back to balance. (post note: following Jan-Feb when, after revisions, we added 88k jobs, in the next 5 months we added only 13k jobs. Jobs growth almost stopped for 5 months.)
So, here we are powering our way through the greatest construction expansion ever recorded, with three years of jobs growth at a 11-year high and jobs growing faster than volume for the past 17 months. Does that seem like a jobs shortage to you?
A few days ago I tweeted, “Last 16 months construction jobs growth outpaced growth in work put-in-place. Hard to see worker shortage from that perspective.”
The imbalance between construction spending and construction jobs is nothing new. It’s been going on for years. It reflects more than just worker shortages. In part it reflects hiring practices. It also captures changes in productivity due to activity. It also helps explain why sometimes new jobs growth rates do not follow directly in step with spending growth. That imbalance can be affected by either over/under-staffing or inflation.
This post presents a series of graphics that show the data that is compared and the percentages of increases (or declines) in balance between workers and output.
What data is available?
We get construction spending from Census and jobs from Bureau of Labor Statistics. For both we can break down the numbers by major construction sectors, Residential, Nonresidential Buildings and Non-building Infrastructure. Inflation is gathered from a number of sources and is specific to sector.
How to look at the data.
Construction spending must be adjusted for inflation to get real volume of work completed. Inflation (or deflation) can vary up or down by 2% to 10% but averages about 4%/year. The adjustment gives us what is referred to as “constant dollars.” Jobs must be adjusted for hours worked. Hours worked can change total workforce output by 2%-3%/year. Everything is converted to the most recent year for comparison.
This Building Cost Index plot shows inflation/deflation by sector. Note that it can vary dramatically from one sector to another. It is often driven by the amount of work activity within the sector. These indexes represent the actual final cost of buildings.
The inflation factors are applied to the annual spending within each sector. This plot shows the combined affect of inflation on total construction output. Current dollars is actual construction spending as reported by U.S. Census. Constant $ is inflation adjusted to 2016. Jobs adjusted for hours worked must be compared to constant dollars.
This plot breaks down the construction spending in constant dollars by sector. This provides the total work completed annually within each sector. This will allow tracking jobs by sector to constant $ by sector.
This plot shows U.S.Census CES jobs by sector. These are adjusted by the hours worked average per year. Hours worked doesn’t change by much within a single year but has varied by 4.5% from maximum to minimum and that has a significant affect on overall output. Adjusted jobs provides us with the total work output.
Those are all the inputs, adjusted to a constant point in time, 2016. Now we can look at how adjusted jobs growth compares to real spending output. I’ve often referred to this as productivity. When spending output is growing faster than jobs, productivity is increasing. When jobs are growing faster than spending output, productivity is decreasing.
The difference between real spending output and jobs growth is more than simply explained as a change in productivity. It reflects the combined impact of hiring practices, hours worked, worker and skills shortages (or excesses) and changes in productivity on inflation adjusted spending. It does not provide a means to differentiate among these causes. However, regardless the cause, imbalances can be thought of as annual productivity gains or losses because they do indeed reflect the total real labor output required to perform the amount of real work put in place.
Leading up to and during the recession there appeared to be far more workers on hand than needed to get the work done. Prior to the recession, I expect a greater portion of the losses were over-hiring and worker productivity losses. It is not uncommon when work is plentiful that productivity declines (2005-2006). When spending started to decrease significantly it took a bit longer for companies to downsize their workforce. During that time (2007-2008) the greater portion of losses might be attributed to insufficient staffing reductions. As we approached the depths of the recession (2009-2010) staffing cuts exceeded the declines in actual work being put in place. Also, people who still had jobs were concerned about keeping their jobs and during such times at first productivity increases. At some point the insufficient staff becomes overworked and productivity declines (2010-2011). Post recession, spending increased faster than companies were replenishing their staff. That led to several years of productivity gains (2012-2015).
Another way to look at this same comparison is to plot the dollars of inflation adjusted volume of work per worker. Here I’ve plotted that by sector.
Some analysts prefer to report this as the number of jobs required to put-in-place $1 billion worth of work. Regardless how it is reported, it is imperative that the comparison be made to constant $, in this case adjusted here to 2016.
Finally we can look at the data for each sector and compare the work being completed each year to the total workforce output to complete that work. There are obviously significant differences in the data by sector.
The data outputs show some things that otherwise are not readily apparent. Note for instance that residential (constant $) spending peaked in 2005, but residential jobs peaked in 2006 when spending was already on the decline. That could be an indication that staffing patterns lag changes in work volume. From the 2005 peak to the 2008 bottom, residential spending declined from $700 billion to $300 billion, almost 60%, but jobs dropped from 3.5 million to 2 million, only about 45%. That would indicate that firms were significantly overstaffed at that time. That explains the several years of deep red bars on the residential productivity plot. That also helps explain in part the slow regrowth on residential jobs. For the real volume of work that was being completed at the time, which was real low, there was already excess staffing remaining on hand.
Work completed and worker output to complete the work will probably continue to be out of balance. It would be difficult to identify any abnormalities in the data releases, for example workers improperly classified to a sector. Any missing workers not captured in the survey would lower productivity. Real productivity gains and losses due to activity will always be a part of the mix. Hiring of less qualified workers due to skilled worker shortages is in the mix. And finally, companies delayed decisions on staffing adjustments will remain part of the issue.