Chapter 20 — Labor, Productivity, and the Construction Workforce

The Hook: Ninety-Two on Tuesday, Sixty-One on Thursday

Thursday, 5:40 p.m., week eighteen. The job trailer smells like burnt coffee and wet gravel. I am looking at two daily reports side by side, and I do not like what they are telling me.

Tuesday: Kestrel's concrete crew placed 92 cubic yards of grade beam and footing concrete. Ten people, ten hours, one pump.

Thursday: same ten people. Same mix design, same supplier, same truck spacing. Same pump, same operator. Sixty-one cubic yards.

That is a thirty-three percent collapse in output over forty-eight hours, on a self-performed activity, on a job where I own the crew, own the schedule, and own the cost code. Nobody quit. Nobody got hurt. The weather was identical — fifty-eight degrees and dry both days.

I walked out to the concrete gang box, where Jamal Foster was writing something on a clipboard with the stub of a carpenter's pencil.

"Jamal. Ninety-two Tuesday, sixty-one today. Same crew. What happened?"

He did not look up. "Pump moved."

"The pump moved."

"Ramp on the east side wasn't backfilled. Excavation sub said Wednesday. It's still Wednesday somewhere, I guess." He finally looked at me. "So we couldn't reach the northwest footings from the east setup like we planned. Set up once, poured the grade beams, tore down, moved, set up again, poured the footings. Forty minutes of setup I didn't have Tuesday. Second washout at the end. And then the part that actually cost you."

"Which is."

"Ten guys on one pour is a machine. Two hoses, two vibrators, a screed gang, a finisher, a guy on the truck, a guy on cleanup. Everybody's got a rhythm and everybody's inside forty feet of everybody else. Split that into two fives and you don't have a machine anymore. You've got two half-machines that both need a full set of tools and a full set of eyes." He shrugged. "Rate went from about eleven and a half yards an hour to about nine and a half. That's not the crew. That's the plan."

Here is what bothers me about this story, and it is not the concrete.

It is that Jamal knew the answer at 10:40 that morning and I did not learn it until 5:40 that evening — and if he had been a subcontractor's foreman instead of my own superintendent, I might never have learned it at all. I would have seen a number on a cost report four weeks later, in a column labeled variance, with no explanation attached, and I would have done what most project managers do with an unexplained labor variance: assumed the crew was slow.

The crew was not slow. The crew was managed into a thirty-three percent loss by a backfill operation on the other side of the site, and every dollar of that loss belonged to me.

Run the arithmetic the way you are about to learn to run it. Kestrel's historical rate for placing and finishing footings and grade beams on this kind of work is 1.15 man-hours per cubic yard. Thursday, the crew earned 61 CY × 1.15 MH/CY = 70 MH and spent 10 workers × 10 hours = 100 MH. The productivity factor for the day was 70 ÷ 100 = 0.70. At a burdened labor rate of $54.00 an hour, the day cost $1,612 more than it should have. One day. One blocked ramp.

There are about forty pour days on Northgate. If that condition — or its cousin, whatever it turns out to be next month — repeats on fourteen of them, it costs $22,568, real money on a job whose entire projected profit is under a million dollars. And nobody will ever write "blocked ramp" on a cost report.

Labor is the only major cost on a construction project that you genuinely control. You do not control steel prices, or what the mechanical subcontractor bid, or the weather, or the owner's decision speed. But you control how many people are on the work face, what they are equipped with, whether the area is ready, what sequence they work in, and how many hours a week you ask of them — and those decisions can swing labor cost by forty percent for reasons that appear nowhere in the contract documents.

This chapter is about making that swing measurable, so you can manage it instead of explaining it.

🏃 Fast Track: If you already run crews, skim 20.1 through 20.3 and go straight to 20.5 (the completed burdened rate, including the effective-rate calculation most estimators skip), 20.6 (the productivity factor and how to measure it in the field), and 20.7 (the overtime arithmetic — the four-week number will surprise you). Do the 📋 Try it in 20.6 with a pencil; do not read the answer first.

🔬 Deep Dive: The burdened rate and the unit-cost formula started in Chapter 12; this chapter finishes them. Productivity measurement becomes formal earned value in Chapter 30 and becomes a claim in Chapter 33. The system fix for most of what you are about to read — making work ready before the crew arrives — is Chapter 27. Productivity reference data and waste factors are in Appendix C; the arithmetic conventions are in Appendix A.

20.1 The One Cost You Actually Control

Recall the estimating identity from Chapter 12. Do not look it up — say it out loud first.

Unit cost = quantity × productivity × rate.

Three terms. Look at what you control over each one after the contract is signed:

Notice how lopsided that is. Two of the three terms are essentially fixed the day you sign, and the third one — the one you own outright — is also the one nobody measures until the cost report arrives.

This is why an estimator can price a job perfectly and a project team can still lose money on it. Tomás Reyes puts 1.15 man-hours per cubic yard in the estimate because that is what Kestrel's crews have historically achieved under normal conditions. Normal conditions are not a law of physics. Normal conditions are a management output. The estimate is a claim that you will produce them; the cost report is the audit of whether you did.

💡 Aha moment. An estimate is a bet on management, not a bet on workers. When a labor cost code goes bad, the first question is never "is the crew any good?" It is "what did we do to the crew's day?" On the overwhelming majority of investigations I have run, the answer was an access problem, a material problem, a sequence problem, or an hours problem — all four of which are decisions somebody in an office made.

There is a second reason labor deserves its own chapter, and it is theme 6 in its bluntest form: you build with people, not with materials. A bar on your Gantt chart is a claim that a specific number of trained human beings will show up on a specific morning, understand a plan they did not write, and cooperate with three other trades standing in the same twelve hundred square feet. Every technique in this chapter is really a technique for making that claim true.

And there is a third reason, which is the one that keeps safety directors awake. The levers that move productivity — hours, pace, crew density, sequence pressure — are the same levers that move injury rates. You cannot push one without moving the other. That is theme 4: safety is not a line item. We will keep coming back to it.

20.2 How the Craft Workforce Is Organized

Before you can budget labor you have to know what you are buying. Construction labor is not undifferentiated "workers." It is a structured occupational system with a ladder, boundaries, and a training pipeline, and every one of those features has cost and schedule consequences.

The craft ladder

Two things about this ladder matter more than the rest.

First, an apprentice is not a discount journeyman. A third-period apprentice carpenter might cost you sixty-five percent of a journeyman's wage and produce sixty percent of a journeyman's output — which means that per unit installed, the apprentice can be more expensive than the journeyman. You do not put apprentices on a crew because they are cheap. You put them on a crew because in four years they are the only reason you still have a crew. That is a company decision with a company payback, and if you evaluate it one job at a time you will always decide wrong.

Second, the foreman is the whole ballgame. A crew foreman decides where eight people stand every morning, whether material was staged the night before, whether the layout is right, and whether anybody says something when a scaffold plank looks wrong. Every dollar of productivity variance in this chapter passes through a foreman. The leadership depth belongs to Chapter 41, but note the economics now: promoting your best installer to foreman removes your best installer from the work face and hands you an untrained supervisor. Do it anyway — but train them first, and know what the first six months cost.

Trades and jurisdiction

The craft workforce is divided into trades: carpenters, laborers, cement masons, ironworkers, operating engineers, electricians, plumbers and pipefitters, sheet metal workers, bricklayers and masons, roofers, glaziers, painters, insulators, elevator constructors, millwrights, and a long tail of specialties. Each has its own training, its own tools, its own hazards, and — importantly for your budget — its own workers' compensation classification code, which is why an hour of ironworker time costs far more than an hour of carpenter time before you even look at the wage. We will price that in 20.5.

Jurisdiction is the question of which trade is entitled to perform a given scope of work. In a union environment, jurisdiction is defined by the international unions' agreements and by decades of decisions, and a jurisdictional dispute — two trades each claiming the same work — is a real schedule risk with a defined resolution process. In an open-shop environment, jurisdiction is whatever your contract and your judgment say it is, which is more flexible and occasionally more chaotic.

Either way, the CM lesson is the same, and it is the scope lesson from Chapter 19: the work between two trades is the work nobody bid. Who sets the sleeve? Who cores the slab? Who hangs the strut the conduit rides on? Who moves the material to the fourth floor? Ask those questions during buyout, in writing, or you will answer them in the field at a premium.

🏗️ From the field. On a hospital job years ago I lost eight days to an argument about who installed the aluminum trim on a nurse-station millwork run — carpenters or the metal trades. Eight days. The trim was worth about four thousand dollars installed. Nobody was being unreasonable; both readings of the scope were defensible, which is exactly what makes these fights expensive. The fix was a fifteen-minute conversation at buyout that nobody had.

Now I run a "who does this?" list at every scope review, and the top of it is always the same six items: sleeves, cores, blocking and backing, hoisting, layout, and cleanup. Those six generate most of the field jurisdiction arguments I have ever seen, in union and open shop alike.

🔄 Check your understanding. Your electrical subcontractor and your drywall subcontractor are arguing about who installs the backing for the wall-mounted panel boards. Why is this a schedule problem before it is a money problem?

Answer

Because the backing has to be installed before the wall is closed, and while they argue, both trades will keep working around it rather than stop. The wall gets closed, the panel board arrives, and now the backing is a demolition-and-repair operation inside a finished wall instead of a two-hour rough-carpentry task — out of sequence, at a rework rate, on the critical path of the finishes. The money problem is a few hundred dollars of blocking; the schedule problem is opening and reclosing walls in a corridor the owner has already walked. Scope gaps between subcontracts become sequence failures in the field.

20.3 Two Labor Systems: Union and Open Shop

Kestrel runs union on public work and open shop on most private work. That is not a philosophical position. It is a response to the price structure that prevailing-wage law creates, and understanding why is a genuine professional skill. Hold these two systems in your head as what they actually are: two different solutions to the same supply problem — getting trained craft workers onto a site that needs them for fourteen weeks and then does not.

The union model

Under a collective bargaining agreement (CBA) — a negotiated contract between a contractor, or a contractors' association, and a local union — the contractor agrees to wage rates, fringe benefit contributions, work rules, and jurisdiction, and in return gets access to the union's labor supply.

• The hiring hall and referrals. You call the hall and request workers by classification. Referral rules vary by local: some operate strict out-of-work lists; most allow some number of "name calls" for specific individuals. A contractor with a good reputation gets good people; one known for short calls and hard treatment gets whoever is left.
• Benefit funds. Health and welfare, pension, annuity, and training contributions are paid per hour worked into jointly trusteed funds rather than your own plan. Those benefits are portable — a journeyman carries them from contractor to contractor — which is one reason craft workers stay in the trade.
• Apprenticeship. Joint apprenticeship and training committees run multi-year programs funded by a per-hour training contribution. You are not building and paying for a training system by yourself. For a contractor with lumpy demand that is a large hidden benefit, chronically under-weighted by people arguing about hourly rates.
• Work rules and jurisdiction. Defined start and quit times, break structure, shift differentials, travel and subsistence, tool provisions, minimum crew compositions in some trades, and defined jurisdictional boundaries. These constrain flexibility. They also remove a great many arguments before they start.
• The project labor agreement (PLA). A PLA is a project-specific pre-hire agreement covering all contractors on one project, setting uniform terms and usually including a no-strike / no-lockout clause and a dispute-resolution procedure. Owners of large or politically visible projects sometimes require one. PLAs are genuinely contested — proponents point to labor peace, uniform terms, and reliable access to trained workforce; opponents point to reduced competition from open-shop bidders and to added cost, and both arguments have real content. If you bid a PLA job as an open-shop contractor, read it closely: you will typically be required to work under its terms and contribute to its funds for the project's duration, which changes your cost structure and sometimes your ability to bring your own core crew.

The open-shop (merit-shop) model

Under an open shop you hire directly. You own the entire problem, and the entire flexibility.

• Direct hire. You recruit, screen, and hire. Your workforce is yours between projects, which means you carry them between projects or you lose them — and the ones you lose are the ones with options.
• In-house training. Some open-shop contractors and associations run excellent structured training and registered apprenticeship programs. Some run nothing and rely on hiring already-trained people — a strategy that works only for as long as somebody else is doing the training.
• Crew flexibility. Broader task assignment across what would be separate jurisdictions under a CBA. On a small job with a five-person crew, that is worth a great deal.
• Retention is your problem. Your differentiator is not a hiring hall; it is being a place people want to keep working. Steady work, predictable schedules, real advancement, and safe sites are your recruiting tools, and every one of them costs money to provide.

The honest comparison

That last row decides the question for most contractors. Which model prevails is heavily regional and sectoral. In some metropolitan areas and in much industrial and heavy-civil work, the union system supplies most of the skilled craft and you cannot staff a large job without it. In other regions, and across most residential work, open shop is simply the market. You do not choose this in the abstract. You choose it where you build, for the kind of work you build.

One more thing, and then we move on, because this is a book about managing projects and not a labor-relations argument. Both models produce superb work and both produce terrible work. The variable that predicts quality and safety is not the model. It is whether there is a real training system, a stable crew, and a supervisor who knows the difference between busy and productive.

⚖️ What the contract says. If your project is covered by a PLA, or if your subcontract flows down a CBA obligation, read three things before signing. One: whether you are bound to the agreement's funds and work rules or only your subcontractors are. Two: the dispute-resolution and no-strike provisions, and what they do to your schedule remedies if work stops anyway. Three: whether the agreement changes your ability to bring existing employees onto the job — because your assumed productivity was built on your own crews, and a crew you have never worked with does not produce your historical rates.

On the subcontract side, your form should require compliance with all applicable labor agreements, prevailing-wage law, and work-authorization requirements, and should give you the right to withhold payment for documented non-compliance. Labor law is heavily jurisdictional and it changes. This is a framework, not legal advice, and the specifics belong to counsel who practices where you build.

🔄 Check your understanding. Kestrel bids Rivermont Elementary School #12 — public, prevailing wage — union, and bids private tenant work open shop. Explain the logic in one sentence.

Answer

On a prevailing-wage job the published wage determination sets a rate everybody must pay, so the open-shop wage advantage largely disappears while the union model's advantages remain — trained craft available on call, shared training cost, and a rate you were going to pay anyway; on private work with no wage determination, the flexibility and rate structure of open shop reassert themselves. It is an arithmetic decision about which system is cheapest to run under the rules that apply to that particular job.

20.4 Prevailing Wage, Certified Payroll, and Classification

If you build anything for a public owner, you will meet prevailing wage. Get this wrong and it is not a variance on a cost report — it is back wages, penalties, withheld payment, and in serious cases debarment from public work, which for many contractors is the end of the business.

What prevailing wage is

A prevailing-wage law requires that workers on a covered public construction contract be paid no less than the wage and fringe benefit rates determined to prevail for their classification in that locality.

• At the federal level, the Davis-Bacon Act covers federal construction contracts above a statutory threshold, and a set of "Davis-Bacon Related Acts" extends similar requirements to many federally assisted projects. The determinations are published by the U.S. Department of Labor.
• Many states have their own prevailing-wage statutes — often called "little Davis-Bacon" acts — covering state and local public work. Some states have none. Thresholds, covered project types, classification systems, apprentice rules, and enforcement all vary, and they change with legislatures and administrations.
• Some local governments add their own requirements on top.

Do not take a rate, a threshold, or a rule out of this book. Get the wage determination that applies to your specific project, for your specific locality and construction type, from the awarding agency and the current published source, and put the date you verified it in your file.

The wage determination and the classification

A wage determination lists, for a locality and a type of construction (building, residential, highway, heavy), each labor classification and two numbers: a base hourly rate and a fringe benefit rate. You must pay the base rate in cash. The fringe portion you may satisfy by contributing to bona fide benefit plans, by paying it as additional cash wages, or by a combination — and you must be able to show which.

Here is an illustrative excerpt in the format you will actually see. These figures are composites for teaching; yours will come from your determination.

Read the spread. A worker setting forms is a Carpenter at $52.05 an hour. The same human being, an hour later, stripping panels and hauling them to the stack, is a Laborer at $39.50. That is a $12.55 per hour difference driven entirely by what the person's hands are doing.

This is where prevailing-wage compliance actually lives, and it is why classification discipline is a field problem, not a payroll-department problem. If a worker performs two classifications in a day, the hours must be split by classification and the split must be supported by records made at the time. "He did carpenter work most of the day" is not a record.

If the work you need does not appear on the determination at all, there is a conformance process to request an additional classification and rate from the contracting agency. Start it early; it is not fast.

Working a certified payroll line

On covered federal work, contractors submit weekly certified payrolls — a payroll report for each week, with a signed statement of compliance by an officer of the company. The certification is not a formality. Under the Copeland "Anti-Kickback" Act, the signer certifies among other things that no rebate or kickback of wages has been taken, and a false certification carries consequences well beyond the contract. State programs have their own forms, portals, and deadlines; many now require electronic submission.

A certified payroll line carries, at minimum: the employee's name and an identifying number, their work classification, hours worked each day and total for the week, the rate paid, gross earnings, itemized deductions, net pay, and how the fringe was satisfied.

Let us work one. D. Alcaraz, carpenter, week ending Friday, on a prevailing-wage job using the determination above. Monday through Friday at eight hours, plus four hours Saturday.

Two nuances worth carrying in your head. First, the overtime premium on covered work is generally computed on the basic rate of pay, and fringe benefit contributions are generally not required to be included in the premium calculation — but the fringe must still be paid for all hours worked, including overtime hours. Second, the Contract Work Hours and Safety Standards Act imposes its own overtime requirement on many covered contracts, separate from the Fair Labor Standards Act. Both of these have wrinkles, both are enforced, and both vary in state programs. Confirm the arithmetic your jurisdiction requires before your first payroll, not after your first audit.

💰 Money check: what one misclassification costs.

Suppose those same 44 hours had been reported as Laborer, Group 1 instead of Carpenter — because it was easier, or because somebody decided form-stripping was laborer work all week, or because nobody was watching.

	Carpenter	Laborer G1
Straight time (40 hr)	$1,394.00 | $1,056.00
Overtime premium hours (4 hr)	$209.10 | $158.40
Fringe (44 hr)	$756.80 | $576.40
Weekly gross	$2,359.90	$1,790.80

Underpayment: $2,359.90 − $1,790.80 = $569.10 — for one worker, for one week.

Now scale it the way an investigator will. Six carpenters, misclassified for twenty weeks:

6 workers × 20 weeks × $569.10 = $68,292 in back wages

That is before interest, before liquidated damages or penalties where the program provides for them, before the cost of the audit, before withheld progress payments, and before the possibility of debarment from public work. On a $22.4 million school job, $68,292 in back wages is survivable. Debarment is not. This is the compliance failure that ends careers, and it almost never starts as fraud — it starts as sloppiness.

Apprentice ratios

Prevailing-wage programs let you pay a registered apprentice at the apprentice rate from their registered program — but only within the ratio the program allows, typically expressed as some number of apprentices per journey-level worker on the job or in the crew. Exceed the ratio and the excess apprentices must be paid the full journey rate for those hours.

Work it. A carpentry crew of eight on a prevailing-wage job: five journeymen and three registered apprentices, under a program allowing one apprentice for every three journey-level workers.

• Allowed apprentices: 5 ÷ 3 = 1.67 → the program permits 1 at apprentice rate (programs differ on rounding — check yours).
• Excess apprentices: 3 − 1 = 2.
• Apprentice package (third period, illustrative): $22.90 base + $12.90 fringe = $35.80.
• Journeyman package: $52.05.
• Shortfall: $52.05 − $35.80 = $16.25 per hour, per excess apprentice.
• For a 40-hour week: 2 × 40 × $16.25 = $1,300 of back wages in one week.

The fix is not complicated. It is a weekly check of the crew composition against the ratio, done by someone who understands both the crew and the rule. On Willow Street that will be you.

⚠️ Safety alert. Ratio compliance has a safety dimension that never appears in the wage discussion. Apprentices are, by definition, the least experienced people on your site, and inexperience concentrates injury risk — insurers and safety researchers consistently find that a disproportionate share of construction injuries occur in a worker's first weeks on a job or in a new task. A crew that is over-weighted with apprentices is not just a wage-compliance problem; it is a supervision problem. If the ratio rule forces you to add journey-level workers, treat that as the safety system doing you a favor.

Worker classification: employee versus independent contractor

Different issue, same family of consequences. Worker classification is the question of whether a person doing work for you is your employee or an independent contractor.

It matters because employees carry obligations that independent contractors do not: payroll tax withholding and the employer's share of FICA, unemployment insurance, workers' compensation coverage, minimum wage and overtime under the Fair Labor Standards Act, and coverage under many benefit and leave laws.

Misclassification is common in construction for structural reasons, and it is worth understanding them without editorializing:

• Work is project-based and intermittent, so a genuine independent-contractor relationship is plausible on its face.
• Multi-tier subcontracting puts distance between the general contractor and the people actually doing the work.
• Some crews are recruited through labor brokers or crew leaders who are paid a lump sum and pay their own people.
• The cost difference is large, which creates competitive pressure on bidders who classify correctly.

The tests that determine status are not uniform. Federal agencies apply different tests for different statutes, and states apply their own — some use a common-law control test, some use an economic-realities test, and some use a strict statutory test. The tests change, and litigation over them is continuous. What is consistent is the direction of the exposure:

Why a general contractor should care about a subcontractor's practices, even though they are not your employees: because under many state statutes and under a growing body of case law, liability for a lower-tier's wage or workers' compensation failure can reach up the chain — and because if a person is injured on your site and turns out to be uninsured, the question of who is responsible for that injury becomes very expensive very quickly. The practical protections are the ordinary ones: prequalify (see Chapter 19), require certificates of insurance and verify workers' compensation coverage is in force before anyone badges in, require certified payrolls where applicable, audit the site roster against the payrolls occasionally, and keep the right to withhold for non-compliance in your subcontract form.

None of that is moralizing. It is risk allocation — theme 1 — applied to the one risk that walks onto your site every morning at 6:30.

🔄 Check your understanding. Why does a prevailing-wage requirement erase most of the open-shop cost advantage but not all of it?

Answer

Because the determination sets the wage and fringe package that everyone must pay for a given classification in that locality, which is the largest single component of the difference. What remains is everything the determination does not set: crew flexibility across jurisdictional lines, work rules and shift structures, the composition of the crew you can field, and your own overhead and administration. Those differences are real but they are much smaller than the wage gap they replace — which is precisely why Kestrel goes union on public work and open shop on private work.

20.5 The Fully Burdened Labor Rate, Completed

Chapter 12 introduced the idea that the wage is not the cost. Here we build the whole thing, line by line, at the level of detail you need to defend a number in front of a CFO or an owner.

The fully burdened labor rate is the total cost to your company of one hour of one worker's time, including everything you pay because that person exists on your payroll.

Here is the complete build-up for a journeyman form carpenter on Kestrel's open-shop private work. Every number is illustrative — yours come from your accountant, your carrier, and your CBA or wage scale — but the structure is universal.

The multiplier: $54.12 ÷ $34.00 = 1.59.

Every dollar you pay a carpenter costs your company $1.59. Tomás carries a flat $54.00 an hour for a form carpenter in the estimating database and reviews it annually; the twelve cents lives inside the rounding.

Why the ironworker costs so much more

Now run the same build for other classifications, and watch line 7 do something dramatic.

Two findings in that table, and both of them are counterintuitive the first time.

Finding one: the ironworker's workers' compensation costs $10.32 an hour against the carpenter's $3.88 — about 2.7 times as much. That is not the insurer being arbitrary. Comp rates are built from the actual injury and claim experience of a classification, and structural steel erection at height generates more severe claims than form carpentry does. The rate is a price signal about hazard, and it is one of the very few places in a construction budget where safety performance shows up as a direct, visible line. Note also the experience modification factor (0.87 here) multiplying every one of those rates: it is your own company's claims history, expressed as a multiplier on your entire payroll-based insurance cost. Cutting your experience mod from 1.00 to 0.87 on Kestrel's self-perform payroll is worth real money on every hour the company works, forever. That is what safety performance is worth in cash — and it is still the smallest of the reasons to do it.

Finding two: the cheapest worker carries the highest burden multiplier. The apprentice at $23.80 base has a 1.71 multiplier while the foreman at $41.00 has 1.54. The reason is that several burden components — health and welfare, small tools, overhead allocation — are flat dollars per person, not percentages of wage. The health plan costs the same for the apprentice as for the foreman. So the lower the wage, the larger those flat costs loom.

The practical consequence: you cannot estimate a mixed crew by applying one multiplier to one average wage. You have to build the crew, price each seat at its own burdened rate, and blend. We do exactly that in 20.8.

The effective rate: the number almost nobody computes

Here is the refinement that separates a good estimator from a great one.

The $54.12 is the cost of an hour on the clock. It is not the cost of an hour on the work. Between those two numbers sits every hour you pay for that produces nothing installed:

If a full-time craft worker is paid for 2,080 hours a year and 294 of them produce nothing installed, then 1,786 hours are actually available for production.

Effective rate = $54.12 × (2,080 ÷ 1,786) = $54.12 × 1.165 = $63.05 per productive hour

That is a 16.5 percent difference, and it is the difference between a labor budget that holds and one that quietly bleeds.

Now — be careful, because this is where people double-count. If your historical unit rates (the 1.15 MH/CY, the 0.035 MH/SF) were derived from actual charged hours on real jobs, then a good deal of that non-productive time is already baked into them, and applying the effective rate on top would be double-counting. The professional discipline is to know which convention your database uses and apply the rate that matches it. Kestrel's rule: unit rates come from charged hours on the cost code, so the estimate uses the $54.12 burdened rate; the effective rate of $63.05 is used only for pricing added scope where the crew has to be pulled off other work.

Write your convention down. Half the estimating arguments I have refereed in twenty-two years were two people using different conventions and each assuming the other used theirs.

🔍 Why this works: the burden is a fixed cost wearing an hourly costume. Most of the burden components — the health plan, the tool allowance, the overhead allocation, the training — are costs the company incurs per person per year, not per hour worked. Dividing them by hours to get an hourly rate is an accounting convenience, and it hides something important: when hours go down, the hourly burden goes up.

That is why laying off a crew during a slow stretch saves less than the arithmetic suggests, and why an idle week is more expensive than it looks. It is also why a contractor with steady backlog can genuinely underbid a contractor with lumpy backlog while paying the same wage — their fixed costs are spread over more productive hours. Backlog is a cost advantage, not just a revenue statistic. That is a company-level insight, and it is where Chapter 34 picks up.

🔄 Check your understanding. An estimator prices added scope by taking Kestrel's historical unit rate of 1.15 MH/CY, multiplying by the effective rate of $63.05, and calling it done. Under what condition is that correct, and under what condition has she just double-counted?

Answer

It is correct only if the 1.15 MH/CY was derived from productive hours — hours actually spent installing. It is double-counting if the 1.15 came from charged hours on a cost code, because charged hours already contain the toolbox talks, the stretch-and-flex, the rain standby, and the moves between areas; the effective rate then adds that non-productive time a second time and inflates the price by roughly 16 percent. Kestrel's rule is explicit: unit rates come from charged hours, so estimates use the $54.12 burdened rate, and the $63.05 effective rate is used only where a crew must be pulled off other productive work. The professional discipline is not knowing one convention — it is knowing which convention your database uses and writing it down. Half the estimating arguments I have refereed were two people using different conventions and each assuming the other used theirs.

20.6 Productivity as a Measurement Discipline

Everything so far has been setup. This is the chapter.

Two measures, and why one is better

They describe the same reality and they convert into each other:

production rate (units per crew-hour) = crew size ÷ unit rate

For Jamal's Tuesday crew: 10 workers ÷ 1.087 MH/CY = 9.2 CY per hour. Same fact, two languages.

Use man-hours per unit as your primary measure. Here is why, and it is not a matter of taste. A production rate is entangled with crew size — change the crew from eight to ten and the production rate changes even if nothing about the crew's efficiency changed at all. The unit rate strips crew size out. It is the measure that stays comparable across crews, across jobs, and across years, which means it is the measure you can build a database from.

And the database is the entire point. Both terms feed straight back into the two documents that run your project:

• Into the estimate: budget man-hours = quantity × unit rate, and labor cost = budget man-hours × burdened rate.
• Into the schedule: duration = quantity ÷ (crew size ÷ unit rate), which is where every duration in your CPM came from whether the scheduler said so or not (see Chapter 14).

That is theme 2 with the math showing: the schedule and the budget are the same conversation, and the unit rate is the sentence they share. Change your assumed productivity and you have simultaneously changed your cost and your duration. Anyone who revises one without revising the other has broken the model.

The productivity factor

The single most useful number in field cost control:

PF = budgeted man-hours earned ÷ actual man-hours spent

where earned man-hours = quantity installed to date × budgeted unit rate.

• PF = 1.00 — performing exactly to the estimate.
• PF > 1.00 — beating the estimate.
• PF < 1.00 — losing.

The genius of it is that it is self-normalizing. It does not care how far along you are, how big the activity is, or what the rate is. A PF of 0.86 means the same thing in week 3 and week 30.

Work it on a real Northgate activity. Slab on grade, cost code 03 3000.30:

Now forecast, which is the part that makes it management information rather than a report card:

Forecast man-hours at completion = budget ÷ PF = 1,155 ÷ 0.927 = 1,246 MH

Check it the other way: 810 ÷ 0.65 = 1,246 MH. Same answer, which is a good habit — a forecast that survives two derivations is a forecast you can put in front of Nadia Haddad.

Overrun = 1,246 − 1,155 = 91 MH Cost impact = 91 MH × $54.12 = $4,925

What it means for the job: you are going to spend about $4,900 more than budgeted on the slab-on-grade cost code, you know it in week 22 instead of finding out at the last pour, and you still have 35 percent of the quantity left to do something about it. That is the whole value proposition of measuring productivity: it converts a post-mortem into a decision.

The earned-value link

If you have peeked ahead to Chapter 30, you have already noticed something. Convert those man-hours to dollars at the budgeted rate:

• Earned value (budgeted cost of work performed): 751 MH × $54.12 = $40,644
• Actual cost of work performed, if the rate held: 810 MH × $54.12 = $43,837
• Cost variance: $40,644 − $43,837 = −$3,193
• Cost performance index: $40,644 ÷ $43,837 = 0.93

The productivity factor is the labor cost performance index — as long as the actual rate equals the budgeted rate. PF is earned value expressed in hours instead of dollars, and hours are the better unit in the field, because a foreman can count hours and quantities but cannot see a dollar.

And when the rates don't match, the difference is itself diagnostic. Suppose those 810 hours included Saturday work, so the actual average rate came in at $57.20 instead of $54.12. Decompose the variance:

Your cost report shows one number: you are $5,688 down. You need two numbers, because the fixes live in different buildings. A rate variance is fixed in the office — stop the overtime, change the crew mix, check whether hours are being charged at the right classification. A productivity variance is fixed in the field — clear the access, stage the material, unstack the trades. Report one number and you will send the wrong person to fix it.

🔄 Check your understanding. Two activities on your job both report a productivity factor of 0.90. Activity A is 15 percent complete; activity B is 85 percent complete. Which one do you spend your Tuesday on, and why?

Answer

Activity A — because there is 85 percent of the work left to influence, and the forecast overrun on it is roughly six times larger than on B for the same PF. Run it: if both had a 2,000 MH budget, A forecasts 2,222 MH with 1,922 hours still to be spent, while B forecasts 2,222 MH with only 333 hours left. The same 0.90 is a live problem on A and a closed book on B. This is the whole reason the productivity factor is worth computing early: a PF is a steering instrument, and steering is only useful while there is road left. On B your job is different and still worth doing — find out why it ran at 0.90 and write it down, because that is a correction to next year's estimating database.

Measuring in the field: the two things you must have

PF requires two inputs, every week, without fail.

One: quantity installed. Not "percent complete" as a foreman's impression. A counted quantity — 610 square feet of contact area formed, 61 cubic yards placed, 340 linear feet of wall framed. This is the daily quantity report, and it lives or dies on whether your foremen believe anyone reads it.

KESTREL CONSTRUCTION GROUP — DAILY QUANTITY REPORT
Job: Northgate Outpatient Pavilion (21-0417)          Date: Thursday, Week 18
Foreman: J. Ruiz                             Superintendent: J. Foster

 COST    DESCRIPTION              UNIT   TODAY   CUM.     BUDGET  CREW  HRS   MH
 CODE                                            TO DATE  QTY
 ------  -----------------------  -----  ------  -------  ------  ----  ---  ----
 031100  Wall / grade beam forms  SFCA      610    9,240  14,800    8    10    80
 033000  Structural concrete      CY         61      398     620   10    10   100
 033000  Slab on grade, place     SF          0    6,900  33,000    0     0     0
                                                                    TOTAL MH  180

 Notes: Pump relocated to NW footing group at 10:40 — east ramp blocked,
        backfill not complete. Crew split 5/5 for balance of pour.
        Second washout required. Lost approx. 1.5 hr of placement time.

That "Notes" block is worth more than the numbers. The number tells you that something happened. The note tells you what. A quantity report without a note field produces a database you cannot diagnose from.

Two: hours charged to the right cost code. This is where most productivity systems die, and it dies quietly. If a worker spends the morning on formwork and the afternoon helping the pour, and all ten hours land on the concrete code, your formwork PF looks great and your concrete PF looks awful, and both are fiction. If a foreman codes everything to one convenient bucket because the code list has 340 entries and no one explained it, you have a payroll system, not a cost system.

Garbage cost codes make productivity measurement impossible. Not hard — impossible. The three disciplines that prevent it, which we develop fully in Chapter 28:

1. Keep the code structure small enough to be used correctly. Twenty-five codes a foreman understands beats three hundred nobody does. Every code should correspond to something a person can count.
2. Match codes to quantities. If a cost code has no unit of measure attached to it, you can never compute a unit rate for it, so it will never tell you anything except how much you spent.
3. Close the loop out loud. When Jamal's report shows a bad day and I ask him about it the next morning, the reporting improves across the whole job. When nobody ever asks, the reports become fiction within a month. Lorena Vasquez, our project accountant, puts it more bluntly: a cost code is a promise between the field and the office, and promises nobody checks stop being kept.

📋 Try it: diagnose a labor cost code.

You are the project engineer on a commercial job. At the month-end cost review you pull one activity — interior metal-stud framing, cost code 09 2216:

Item	Value
Budget man-hours	1,850 MH
Quantity installed to date	62% of the total quantity
Actual man-hours charged to date	1,340 MH
Fully burdened labor rate	$54.00/hr

Do all five, on paper, before opening the answer:

1. Compute the productivity factor.
2. Forecast the man-hours at completion.
3. Compute the variance in man-hours and in dollars.
4. Compute the productivity factor the remaining work would have to achieve for the activity to finish on budget — and say whether that is realistic.
5. Name the three causes you would investigate first, in order, and say what evidence would confirm or eliminate each.

Worked answer

1. Productivity factor.

Earned MH = 62% × 1,850 = 1,147 MH PF = earned ÷ actual = 1,147 ÷ 1,340 = 0.856 → PF = 0.86

You are getting about 86 cents of budgeted work out of every hour you buy.

2. Forecast at completion.

Forecast MH = budget ÷ PF = 1,850 ÷ 0.856 = 2,161 MH

Cross-check: actual ÷ percent complete = 1,340 ÷ 0.62 = 2,161 MH. Two derivations, same answer.

3. Variance.

Variance in hours = 2,161 − 1,850 = 311 MH over budget Variance in dollars = 311 MH × $54.00 = $16,794

Put it in plain English for the owner of this cost code: at today's rate of production, this activity finishes about 311 man-hours and roughly $16,800 over budget.

4. What the remaining work would have to do.

Hours remaining if you finish on budget: 1,850 − 1,340 = 510 MH Work remaining, in earned hours: 38% × 1,850 = 703 MH Required PF for the remainder: 703 ÷ 510 = 1.38

The rest of the job would have to run at 38 percent better than the estimate — not better than current performance, better than the estimate, which was already built on good conditions. That is not realistic. This activity cannot be recovered; it can only be contained. Learning that in the middle instead of at the end is exactly the point, because containment still saves real money: if you can lift the remainder from 0.86 to 0.95, the forecast drops from 2,161 to about 1,340 + (703 ÷ 0.95) = 2,080 MH — roughly 81 man-hours and $4,400 saved, on work you had already lost.

5. The three causes to investigate first.

(a) Measurement integrity — is 62 percent real, and are the hours real? Investigate this first, always, because if the inputs are wrong every other conclusion is wrong. A foreman's optimistic percentage, or hours from an adjacent activity (in-wall blocking, ceiling grid, patching) landing on the framing code, will fabricate a variance out of nothing. Evidence: recount the quantity physically — count studs or measure linear feet of wall by area, do not accept a percentage; then pull the individual time cards for two random weeks and ask the foreman what each person actually did.

(b) Access and interference — did the work environment change? Framing is early in the interior sequence and gets stacked on fast. Evidence: compare the areas being framed to the six-week look-ahead; walk the floor at 9:30 a.m. and count how many trades are in the same room; check whether framing is following the planned area sequence or jumping around because areas were not released. Out-of-sequence framing carries setup cost per move that the unit rate never assumed.

(c) Crew composition and hours. Evidence: compare the actual crew mix and headcount on the daily reports against the mix the estimate assumed, and pull the overtime hours. A crew that grew from eight to twelve on the same work face, or a crew that lost two journeymen and gained two apprentices, or a crew that has been on 10-hour Saturdays for five weeks, will produce exactly this signature.

Honorable mentions if all three come back clean: rework (check the nonconformance and punch logs), material shortages or wrong-length stud deliveries (check the delivery log and the layout), and layout errors requiring re-framing (check the RFI log for dimensional questions in these areas).

The meta-lesson: notice that only one of these three candidate causes is about how hard people are working, and it is not the first one you check.

Work sampling and the five-minute rating

PF tells you that productivity is off. It does not tell you where the day is going. For that you observe.

The five-minute rating is the field version, and any superintendent can do it. Pick a crew. Watch for at least five minutes. At fixed instants — say every thirty seconds — record how many crew members are actively working. The rating is:

rating = working observations ÷ total possible observations

Six people watched at ten instants gives 60 possible observations; if 33 of them show someone working, the rating is 55 percent. You are not judging speed and you are not judging people. You are counting a state at an instant, which is the only observation cheap enough to take often and honest enough to be worth taking.

Work sampling (also called activity analysis) is the same idea run larger: a trained observer takes hundreds of instantaneous observations across a site on a random route, classifying each into categories. Here is Northgate, week 30, level 2 interiors, 1,212 observations across all trades in the area:

Studies of construction work sampling commonly report direct work somewhere in the range of roughly a third to a half of the workday, with everything else distributed across the other categories. Treat any specific published percentage with suspicion, including this one — the category definitions differ between studies, what counts as "direct" differs, and the number varies enormously by trade, by phase, and by building type. Chasing an industry benchmark is a waste of your time.

The useful finding is never the direct-work number. It is which non-direct category is biggest on your job, because each one points at a different management failure:

On Northgate that week, travel plus waiting was 19% + 12% = 31% of every observed instant. Nearly a third of the workday, and not one minute of it was a worker's decision.

Two rules about using this, and I want you to take both seriously.

Rule one: never use work sampling as a stick. The instant a crew believes the clipboard is looking for slackers, three things happen: the observations become theater, the foremen stop cooperating, and you lose the only cheap diagnostic tool you had. Announce what you are doing and why. Share the results with the foremen first, before anyone in the office sees them. Frame every finding as a question about the work environment, because that is what it is measuring.

Rule two: do not convert a sample into a claim. A work-sampling study is a management diagnostic, not a forensic proof of disruption damages. When you need to prove lost productivity in a dispute, the defensible approach is a measured mile — comparing your own performance in an unimpacted period against the impacted period on the same work — and that is Chapter 33's territory.

🧩 Productive struggle. Stop here and work this before reading 20.7.

Margo Deacon needs the level-2 deck-edge formwork finished in three ten-hour days instead of four. There are 2,720 SFCA remaining. Your crew of eight produces 680 SFCA per ten-hour day. You need 907 a day.

Write down: (a) how you would get there, (b) roughly what you think it costs, and (c) which of your options you think is cheapest per day recovered.

Take three minutes. Do not skip this — the arithmetic in 20.7 and 20.8 will land much harder if you have committed to an answer first. The honest solution uses both sections, and at least one of the three obvious options does not work at all.

20.7 What Actually Moves Productivity

Every factor below has a mechanism — a physical or organizational reason it changes output. Learn the mechanism, not the number. The numbers are planning heuristics; the mechanisms are how you diagnose a job you have never seen before.

Overtime and the fatigue curve

This is the big one, it is the most misunderstood, and it is the one that will cost you the most money.

Everyone knows overtime costs a premium. Almost nobody prices the efficiency loss, and on a sustained schedule the efficiency loss is larger than the premium.

First, the premium, computed correctly. People routinely price an overtime hour at 1.5 times the burdened rate. That is wrong, and it is wrong in the expensive direction — it makes overtime look worse than it is, right up until the efficiency loss makes it far worse than they thought.

The half-time premium applies to the base wage, not to the burdened rate, because health and welfare, tools, and overhead allocation do not repeat with the extra hour. The statutory burdens (FICA, unemployment, workers' compensation, general liability) do apply to the premium, because it is wages:

Statutory loading on wages = 7.65% + 1.10% + 10.88% + 1.65% = 21.28% Premium half-hour = $34.00 × 0.5 = $17.00 Loaded premium = $17.00 × 1.2128 = $20.62 Cost of one overtime hour = $54.12 + $20.62 = $74.74

That is 1.38 times straight time, not 1.5. Pricing it at 1.5 × $54.12 = $81.18 overstates each overtime hour by about $6.44. Your accountant will have a convention; use theirs, use it consistently, and make sure it is not the naive one.

Second, the efficiency loss. The pattern is old, well documented, and consistent across many sources: when a crew works sustained overtime — more than 40 hours a week, week after week — output per hour degrades. It degrades progressively. It is small in week one and substantial by week four or five. The mechanisms are not mysterious:

• Physical fatigue, cumulative and not fully recovered by a short weekend.
• Absenteeism rises, because people schedule around a six-day week however they can, and a crew missing two people does not lose two people's output — it loses its rhythm.
• The pace self-adjusts. A crew that knows it is working sixty hours settles into a sixty-hour pace. This is not malingering; it is how humans allocate effort across a long day.
• Support functions do not scale. The layout, the material deliveries, the inspections, the crane, and the foreman's attention were sized for a forty-hour operation. The extra twenty hours are worse-supported hours.
• Error and rework rise, which shows up on a different cost code entirely and is therefore invisible in the overtime discussion.

The mechanical and electrical contracting associations publish factor tables for this; the Business Roundtable studied it in a well-known series in the 1980s; the Corps of Engineers and various claims authorities have published guidance. They differ in the details and they agree on the shape. I am not going to give you a citation-grade number, because the honest answer is that it depends on the trade, the work, the weather, the crew, and how long the schedule runs. What follows are illustrative planning factors — the kind you would use to price a risk, not to prove a claim.

💰 Money check: the true cost of a 10×6 schedule.

Crew of 10 form carpenters. Four weeks of ten-hour days, six days a week — 60 hours a week. Straight-time burdened rate $54.00; overtime hour $74.60 (using round numbers).

Step 1 — the weekly cost.

	Hours	Rate	Cost
Straight time	40	$54.00 | $2,160
Overtime	20	$74.60 | $1,492
Per worker per week	60		$3,652
Crew of 10 per week	600		$36,520

Baseline 40-hour week: 10 × 40 × $54.00 = $21,600. So you are buying 50% more hours for 69% more money. Most people stop here. Do not stop here.

Step 2 — apply the efficiency curve. Illustrative planning factors: week 1 at 1.00, week 2 at 0.95, week 3 at 0.90, week 4 at 0.85.

Week	Hours worked	Efficiency	Effective (productive-equivalent) hours
1	600	1.00	600
2	600	0.95	570
3	600	0.90	540
4	600	0.85	510
Total	2,400		2,220

Four weeks at 40 hours would have given 4 × 400 = 1,600 effective hours for 4 × $21,600 = $86,400. Four weeks at 10×6 gives 2,220 effective hours for $146,080.

Step 3 — the number that matters: cost per unit of real output.

40-hour baseline: $86,400 ÷ 1,600 = $54.00 per effective hour 10×6 schedule: $146,080 ÷ 2,220 = $65.80 per effective hour

Every unit of actual production costs 21.9 percent more.

Step 4 — the marginal cost, which is the number to quote in the meeting. You did not buy 2,220 hours. You bought 620 extra effective hours (2,220 − 1,600) for $59,680 extra ($146,080 − $86,400):

$59,680 ÷ 620 = $96.26 per extra effective hour — 1.78 times straight time.

And it gets worse as it runs. Take week four by itself: $36,520 buys 510 effective hours where a normal week's $21,600 buys 400. The extra 110 effective hours cost $14,920:

$14,920 ÷ 110 = $135.64 per extra effective hour — 2.51 times straight time.

By week four you are paying two and a half times the straight-time rate for every additional hour of real production, and you are also paying in near-misses, rework, and turnover, none of which appear on this table.

This is the arithmetic behind the canonical Northgate acceleration. Kestrel priced the recovery package at $168,000 for 17 days. What that package did not price is what case-study-02.md measures.

The practical rules that come out of that arithmetic:

1. Short bursts of overtime are efficient. Sustained overtime is not. One or two weeks to hit a pour date or a dry-in milestone is a good tool. Ten weeks is a way of spending money to go slower.
2. Compressing a week is not the same as extending it. Three twelve-hour days is thirty-six hours — no weekly overtime premium at all in most jurisdictions. Keep that in your pocket; we use it in 20.8.
3. Before you buy overtime, buy crew size, sequence, and access — they are almost always cheaper per recovered day.
4. If you must run sustained overtime, rotate it. Two crews at fifty hours beats one crew at sixty, and it is safer.
5. Overtime hours are a leading indicator. Track overtime as a percentage of total hours weekly. A cost code drifting above ten percent overtime without a written reason is telling you something before the cost report does.

⚠️ Safety alert: the fatigue-injury link. Everything in that money check has a shadow on the safety side, and the shadow is the reason theme 4 exists.

Fatigue degrades exactly what construction work depends on — reaction time, balance, judgment about risk, and the willingness to stop and fix something. Long shifts and consecutive days without rest are consistently associated with higher injury rates; this is one of the most robust findings in occupational safety research across industries, and construction has no exemption from it.

New workers are at concentrated risk. A disproportionate share of construction injuries occurs in a worker's first days and weeks on a job or in a new task. When you accelerate by adding people, you are adding exactly the population most likely to get hurt, into an environment that is now more congested and moving faster, often on a Saturday when the safety staff is thinner. Orientation is not paperwork; on an accelerated job it is the highest-value hour you will spend all week.

Heat illness kills, and it kills early. Heat-related deaths in construction concentrate in workers' first days of hot work, before the body has acclimatized. Water, rest, shade, and a real acclimatization schedule for new and returning workers are not comfort measures. OSHA has pursued heat hazards under the General Duty Clause, several states have their own heat standards, and NIOSH publishes the underlying criteria. If you push a 10×6 schedule through July, you have made a heat-illness decision whether or not you know it.

And the one that indicts management directly. Go back to the scaffold near-miss in week 34 — Milo Serrano, north elevation, level 3, a plank lifted overnight by another trade and not re-secured. Bea Salgado's investigation found three failures. The first two were the missing inspection tag and the unauthorized modification. The third was that the crew was running behind after the steel acceleration, under an unwritten expectation to make it up. That expectation was not in any document. It came from the schedule pressure created in this chapter's arithmetic.

Schedule pressure is a hazard exactly like an unguarded edge — with the difference that an unguarded edge is created by a worker and schedule pressure is created by a project manager. Chapter 24 takes the whole system apart.

The rest of the factor list

Trade stacking and congestion. More than one trade working in the same physical area at the same time. The mechanism is interference: people wait for each other's ladders, lifts, and cords; material staging competes for the same floor; and every trade's setup gets moved by somebody else. This is Chapter 19's coordination problem showing up as a labor number, and it is the single most common cause of a bad PF on interior work.

Out-of-sequence work and rework. Work performed out of its planned order carries a setup cost per move that the unit rate never contemplated. Rework is worse still, because you pay three times — once to build it, once to remove it, once to rebuild it — and because the removal usually happens in a more finished, more congested condition than the original. Track rework on its own cost code. If you bury it inside the production code, you will never know what your quality problems cost.

The learning curve. On repetitive work, unit rates improve at a fairly predictable rate as cumulative output doubles. This is a real, exploitable effect, and Harbor Ridge is built on it. Colton Reyes runs eleven houses at a time on a 92-calendar-day per-house cycle for Tessa Bright Homes, and the reason that cycle holds is repetition.

Model it as a 90 percent curve — each doubling of cumulative units takes 90 percent of the previous doubling's hours. Framing labor per house:

By the thirty-second house, framing takes 41 percent fewer hours than the first. Across all 34 houses the model predicts roughly 15,000 man-hours of framing against about 21,800 if nothing were learned — call it 6,800 man-hours, and at a burdened residential framing rate near $46 an hour, roughly $310,000. That is the entire margin on several houses.

Which is exactly why Colton will not let a framing subcontractor rotate crews in from another subdivision mid-run. Crew continuity is a schedule and cost asset with a dollar value, and most people give it away for free. The same logic applies to a commercial job with repetitive floors: your fourth floor should be measurably faster than your first, and if it is not, something is wrong with your sequence, not with your crew. See Chapter 37 for the residential production system in full.

Weather and temperature. Cold costs dexterity, bulky personal protective equipment, protection and heating operations, and concrete and masonry restrictions. Heat costs work/rest cycles, hydration breaks, and — above roughly 90°F with humidity — a real and non-negotiable slowdown you should plan rather than fight. Wind shuts down cranes and anything at height. Rain costs the day plus the two hours of cleanup after it.

Height above grade and travel time. The mechanism is vertical transport. Every floor between the gate and the work face costs walking time twice a day, plus every trip for material, plus hoist queueing. A crew on level 8 with one hoist is not the crew it was on level 2.

Shift work and night work. Second and third shifts are legitimate tools — occupied buildings, street work, tie-ins that require an outage. They cost productivity through thinner supervision, artificial lighting, communication gaps between shifts, and circadian disruption, and they cost a shift differential on top. Plan a shift overlap for handoff; twenty minutes costs less than the confusion it prevents.

Access and material availability. The most common and most fixable loss on any job. If material is at grade and the work is on three, somebody is carrying it. If the layout is not shot, somebody is measuring. This is where Chapter 18 meets Chapter 27, and it is where the largest recoverable losses on most jobs sit in plain sight.

Crew size and the optimum crew. Both too small and too large hurt. That is 20.8.

Supervision ratio. One working foreman can effectively direct roughly ten to fifteen people on similar work in one area — fewer if the work is complex, spread out, or hazardous. Beyond that, direction lags: people wait for answers, layout errors survive longer, and the foreman stops working with their hands, which on a small crew is often the pace-setting function.

Absenteeism and turnover. Absenteeism is not linear. An eight-person crew at 92 percent attendance is not 8 percent slower — if the missing person is the finisher or the welder, the crew is stopped. Turnover costs more than most contractors ever measure:

A forty-person self-perform workforce at 35 percent annual turnover means fourteen replacements a year: 14 × $4,600 = $64,400 — every year, invisibly, in a line item nobody has. And the safety cost rides along with it, because those fourteen people are all new.

The productivity adjustment factor table

Use this to price risk in an estimate and to frame a diagnosis in the field. Multiply the baseline unit rate by the factor.

Read these caveats before you use that table, and read them every time.

1. These are planning heuristics, not measured constants. They are for pricing risk and framing conversations. They vary by trade, by market, by building type, and by crew.
2. They do not compound cleanly. Multiplying five factors together produces arithmetic nonsense — a factor of 2.8 that no honest analyst would defend. Where multiple conditions apply, take the dominant one and add a judgment increment, and say in writing that is what you did.
3. They are not a claims methodology. A schedule of factors applied to a disruption claim is the weakest form of proof there is, and sophisticated owners' consultants dismantle it routinely. When money is genuinely at stake, you need contemporaneous records and a measured-mile analysis — Chapter 33.
4. The best table in the world is the one you build from your own cost history. Every job you finish is a data point. A contractor with ten years of clean cost codes has an estimating advantage no published reference can match, and building that database is the highest-return administrative work in a construction company.

🔄 Check your understanding. Your superintendent proposes going to Saturdays for the next eight weeks to recover four days. Before you approve it, what are the three numbers you need?

Answer

One: the overtime premium cost — the extra dollars per hour, computed on the base wage with statutory loading, not naively at 1.5 times the burdened rate. Two: the efficiency loss over the eight-week duration, which on a sustained schedule will likely exceed the premium and which determines how many days you will actually recover, not how many you hope to. Three: the daily value of the days recovered — on Northgate, $10,650 per calendar day of extended general conditions plus liquidated damages. Without the third number you cannot tell whether the first two are worth spending. And a fourth you should ask for even though it is not a number: what does eight weeks of six-day weeks do to your incident rate and your turnover, and is there a cheaper way to buy those four days — sequence, access, or crew size?

20.8 Crew Composition, Balance, and the Arithmetic of Adding a Person

A crew is not a headcount. It is a designed thing, with a mix of skills, a rhythm, and an optimum size for a particular work face. Designing one is a real engineering exercise, and getting it wrong is the most common self-inflicted productivity wound in the business.

Designing a crew

Three questions, in order.

1. What does the work actually require? Break the operation into its steps and staff each one. Wall formwork with gang panels needs: layout, panel setting, aligning and plumbing, walers and ties, bracing, and material handling and cleanup. Layout and alignment are journeyman work. Panel setting is journeyman-plus-apprentice. Material handling and cleanup is laborer work.

2. What is the balance — is any step starving another? An unbalanced crew is one where a step upstream cannot feed the step downstream. Four journeymen setting panels faster than two laborers can bring them is an unbalanced crew, and the symptom is journeymen carrying plywood. You are paying $54.12 an hour for work you could buy at $41.51, and you are also losing the panel-setting production those hands were supposed to be doing. The classic tell: your highest-paid people are doing your lowest-paid work.

3. What does the work face physically hold? Every work face has a ceiling. Twelve people cannot stand on a forty-foot wall section. When you hit that ceiling, more bodies produce zero and cost full price.

The optimum crew, worked

Wall and grade-beam formwork on Northgate, ten-hour days. Burdened rates from 20.5: foreman $63.32, journeyman $54.12, apprentice $40.70, laborer $41.51.

The optimum is nine people at $5.70 per square foot of contact area. That is the crew Jamal builds when nobody is pushing him.

Now the part that changes how you think. Look at the marginal cost — what each additional person actually buys:

The ninth person delivers formwork at $4.07 a square foot against a budget of about $5.90. Buy that person every time.

The tenth person delivers formwork at $9.02 a square foot — more than fifty percent above the budgeted unit cost. The eleventh delivers it at $20.76, more than three times budget.

And here is what makes this so dangerous in practice: all eleven of those people look busy. Walk the wall at ten in the morning and you will see eleven people working. The loss is invisible to observation and completely visible in arithmetic, which is precisely why you measure instead of eyeballing.

🔍 Why this works: what the curve is actually made of. The unit-rate curve is U-shaped for two independent reasons pulling in opposite directions.

Going up from six people, you gain specialization. At six, the journeymen are handling their own material and doing their own cleanup — high-wage hands doing low-wage work. Each person you add lets everyone else stay in their own lane, and the whole crew speeds up more than the added person's own output.

Past the optimum, you hit interference. The work face is a fixed length of wall. Adding people means shorter reaches, more waiting for a hand, more people crossing paths, and more time coordinating instead of installing. The added person's own production is real but small, and they slow everyone else down a little.

The optimum is where those two forces cross. It moves — a longer wall pushes it right, a congested corner pushes it left, a hoist bottleneck pushes it hard left. This is why "how many people should be on this?" has no general answer and always has a specific one. Ask your foreman. They know. Almost nobody asks.

Now solve the productive struggle

Back to Margo's deck edge: 2,720 SFCA in three ten-hour days, needing 907 SFCA per day against a crew of eight producing 680.

Option A — bodies alone. The curve says this work face tops out around eleven people at 860 SFCA a day. You need 907. Bodies cannot do it. That is the first lesson, and it is the one most people miss: the work face has a physical ceiling, and past it you are buying nothing at full price.

Option B — hours alone. Crew of eight on twelve-hour days: 680 × 12/10 = 816 nominal, less something for fatigue and the extra setup and cleanup at each end — call it 780 a day. Still short of 907. Hours alone cannot do it either.

Option C — both. Crew of ten on twelve-hour days: 840 × 12/10 = 1,008 nominal, times roughly 0.93 for the long day, gives about 937 SFCA per day. 2,720 ÷ 937 = 2.9 days. That works.

Price it:

Extra cost = $17,940 − $16,141 = $1,799 to buy one day.

Three things in that table are worth more than the answer.

First, there is no overtime premium in it at all. Three twelve-hour days is thirty-six hours — under forty. Compressing a week is not the same as extending it, and the difference is the entire premium. (Caveat, and it is not a small one: some states require daily overtime after eight or ten hours regardless of the weekly total, and many collective bargaining agreements do the same. Check your jurisdiction and your agreement before you use this move.)

Second, the blended rate went down when the crew got bigger — $50.44 to $49.83 — because the two people added were an apprentice and a journeyman rather than two journeymen. Crew mix is a real lever on the rate term of the unit-cost formula, and it is the only lever on that term you control mid-job.

Third, and this is the point of the whole chapter: $1,799 to buy a day, on a job where a calendar day is worth $10,650, is one of the best trades available to you. The cheap days are bought early, with sequence and crew design. The expensive days are bought in week 34, with a sixty-hour schedule, at $135 an hour of real production.

🔄 Check your understanding. Your foreman says he needs two more people. What do you ask him before you say yes?

Answer

Four questions. (1) What are they going to do? — if the honest answer is "help," you have an access or material problem, not a headcount problem. (2) What will the daily production be with them, and what is it now? — that gives you the marginal production, which is the only number that matters. (3) What classification? — adding two laborers to free journeymen from material handling is usually a good buy; adding two journeymen to a crowded face usually is not. (4) Is the work face big enough to hold them? Then do the marginal arithmetic: extra crew cost per day divided by extra production per day, compared against the budgeted unit cost. If the answer is above budget, you may still buy them — but now you are buying schedule, deliberately, at a known price, and you can say so in writing.

20.9 The Workforce Problem, Told Straight

Back in Chapter 2 I said that the labor supply is going to be the binding constraint on this industry for the next decade, and that it is a schedule problem before it is a cost problem. Everything in this chapter has been building the tools; here is the problem those tools are for.

The craft workforce is aging. The average age of skilled craftworkers has been drifting upward for years. The Bureau of Labor Statistics publishes the demographics and the direction is not in dispute. A great many highly skilled people are within a decade of retiring, and they carry knowledge that lives nowhere else — nobody has written down how Jamal Foster reads a formwork joint or how Margo Deacon knows from the sound of a job at seven in the morning that something is wrong.

The pipeline is thinner than the demand. Two generations of policy and culture pushed young people toward four-year degrees and away from the trades, and a great many high schools eliminated shop programs entirely. That is partly reversing — career and technical education has been rebuilding, and both union joint apprenticeship programs and open-shop programs run through associations like AGC and ABC are real and effective. They have not been scaled to the size of the gap, and the shortage worsens exactly when construction demand is strongest, which is the least convenient possible timing.

Retention is as big a problem as recruitment, and it gets a fraction of the attention. The industry loses a meaningful share of the people it recruits within their first years, and the reasons people give are consistent: unpredictable schedules and layoffs, travel, the physical toll, a lack of a visible path forward, and treatment. Every one of those is addressable, and none of them is addressed by a recruiting campaign.

Now the part that gets handled badly, so let us handle it plainly. Women are a small minority of the craft workforce — a low single-digit percentage of the people actually working with tools, a figure that has barely moved in decades. That is not a mystery and it is not a preference. The barriers that recruiters and researchers identify most consistently are structural and fixable:

Look at that column of costs. Almost every item is either free or a few hundred dollars. This is not a values argument; it is a supply argument. An industry with a labor shortage that recruits effectively from something under half the population has diagnosed its own problem. The same reasoning applies to every under-recruited population — veterans, people re-entering after incarceration, career changers, and workers whose first language is not English and who are consistently underserved by training material that exists only in English.

What actually works is unglamorous and well established: real apprenticeship — structured, paid, classroom instruction, a credential at the end, in both union and open-shop forms; career-path clarity, meaning a person can see on one page what it takes to go from helper to apprentice to journeyman to foreman, how long each step takes and what it pays (most companies cannot produce this page; producing it costs an afternoon); predictable schedules, which cost you backlog management you should be doing anyway; safety, because people do not stay where they watch others get hurt and they tell their friends — your incident rate is a recruiting statistic; respect expressed as process — being asked before the plan is set, a functioning gang box and a clean job, an answer within a day, not being humiliated in front of a crew, all free and all rare; and pay, honestly stated, which matters and is not the only thing that matters. Treating rate as the only lever is why some contractors pay top scale and still cannot keep anybody.

Productivity is a management variable

Which brings the chapter back to where it started. Look back at every factor in 20.7 and ask: who decided that?

Sequence — the project team. Access — the project team. Material staging — the project team. Crew size — the superintendent. Overtime — the project manager. Trade stacking — the coordination process. Supervision ratio — the company. Rework — usually the coordination or the information flow. Weather — nobody, but planning for weather is the project team.

The crew's own effort is somewhere on that list, and it is nowhere near the top.

That is the single most important idea in this chapter, and it is theme 6 stated as arithmetic: productivity is a management variable more than it is a worker variable. When you understand that, you stop asking your foremen to push and start asking them what is in the way. Those two conversations produce very different numbers, and only one of them produces a workforce that is still there next year.

The leadership craft that grows out of this — how to run a crew, how to handle conflict, how to be the kind of manager people follow into a hard week — is Chapter 41.

🔄 Check your understanding. Your company's craft turnover is 35 percent a year and your recruiting budget just doubled. Explain, with a number, why that is probably the wrong lever.

Answer

Because recruiting refills a bucket you have not stopped leaking, and every refill costs about $4,600 in orientation, ramp-up, and foreman training time before the new person is worth their rate — fourteen replacements a year on a forty-person craft workforce is roughly $64,400, plus a safety cost that never appears on any report, since injuries concentrate in a worker's first weeks. Reducing turnover from 35 percent to 20 percent saves six replacements a year, about $27,600, and raises average crew productivity, because a stable crew is further along its learning curve. The retention levers — predictable schedules, a visible career path, safety, and respect expressed as process — are mostly cheap or free. Recruiting harder into a leaking bucket is the most expensive way to solve a retention problem, and it is the one most companies choose because it feels like action.

Where the ethical lines are

Every technique in this chapter can be misused, and the misuses are common enough that you will see all four before you are thirty-five. Here is the line, and why crossing it costs more than it pays.

Wage theft. Shaving time from cards, requiring off-the-clock work before or after the shift, paying "straight time for overtime," misusing per diem as a substitute for wages, or paying part of a check in cash to avoid the burden. Every one of these is illegal, all of them are discoverable in an audit, and they carry back wages, penalties, and in serious cases criminal exposure. They also destroy the thing that makes a crew productive — a crew that thinks you are shorting them is not going to tell you when the plank is loose.

Misclassification. Covered in 20.4. It is often inherited rather than chosen — you did not classify anyone, your labor broker did — which is exactly why you have to look.

Pressuring a crew to "make it up." This is the subtlest one and the most damaging, because it never gets written down and it feels like leadership. "I need this deck by Friday, I don't care how." No directive, no plan, no added resources, no acknowledgment that the schedule slipped for reasons the crew did not cause. What you have actually communicated is that the date matters more than the method — and the method is where the safety is. Finding #3 in the scaffold investigation is exactly this. If you need a date, provide a plan and resources for it. If you cannot, then you need a different date, and the honest conversation is with the owner, not with a foreman.

Safety incentive programs that suppress reporting. A program that pays a bonus for an injury-free month sounds like it buys safety. What it frequently buys is silence — an injured worker under peer pressure not to cost the crew its pizza party, a foreman who finds a reason not to record something. OSHA's recordkeeping rule includes anti-retaliation provisions, and OSHA has issued guidance making clear that incentive programs which discourage reporting can themselves be a problem. The fix is to incentivize leading indicators — near-miss reports submitted, hazards corrected, inspections completed, toolbox talks held — never lagging ones. A crew that reports fifty near-misses is a healthy crew. A crew that reports zero of anything is a crew that has learned not to speak.

Spaced Review

Cover the answers. Produce each one from memory first — the effort of retrieval is what makes it stick, and it will not feel comfortable, which is the point.

1. From Chapter 19 — the manpower curve as an early warning. What is a manpower loading curve, and why does it tell you a subcontractor will miss a date before the schedule update does?

Recall first. — A manpower loading curve plots planned workers on site by week for a trade against actual counted workers. The reason it leads the schedule update is arithmetic: a subcontractor's duration was built on an assumed crew size, so if they planned twenty and staffed twelve, the duration has already stretched by roughly two-thirds even though no activity has reported late yet. The schedule update tells you about a date that has already slipped; the manpower curve tells you about a date that is going to slip, four to six weeks earlier, while there is still time to do something. Northgate:

 NORTHGATE — TOTAL CRAFT MANPOWER, FRIDAY HEAD COUNT
                PLANNED  ACTUAL   ACTUAL (# = 10 workers)         GAP
   Week 20         78      74     #######                          -4
   Week 24         96      88     #########                        -8
   Week 28        118     101     ##########                      -17
   Week 32        142     116     ############                    -26
   Week 36        166     129     #############                   -37
   Week 40        182     158     ################                -24

The gap widens for four consecutive updates before it closes. Week 28 is where you make the phone call, not week 36. And note the scale of the whole thing: Northgate's peak craft workforce is 210 workers in week 61 against roughly 412,000 total craft man-hours over 565 calendar days — about 80 work weeks, an average near 5,150 man-hours a week, or roughly 129 people at forty hours. Peak-to-average of about 1.6 is normal, and it is the entire reason site logistics is hard: you must build parking, sanitation, hoisting, and laydown for 210 people on a site whose average population is 129.

2. From Chapter 12 — the burdened rate and the unit-cost formula. State the unit-cost formula, then explain why the wage is not the rate.

Recall first. — Unit cost = quantity × productivity × rate. The wage is not the rate because a worker costs you far more than their wage: paid time off, health and welfare, retirement, FICA, unemployment insurance, workers' compensation by class code, general liability allocated on payroll, small tools, and overhead. On Kestrel's journeyman carpenter that build-up runs $34.00 to $54.12, a 1.59 multiplier. Two refinements this chapter added that Chapter 12 did not reach: the multiplier is higher for lower-wage classifications because several burden components are flat dollars per person; and the on-the-clock rate is not the on-the-work rate — dividing by productive hours gives an effective rate near $63.05, which you must use only if your unit rates do not already contain the non-productive time.

3. Deep callback to Chapter 2 — the aging craft workforce. Why is a craft labor shortage a schedule problem before it is a cost problem?

Recall first. — Because a shortage does not announce itself as a wage increase you can forecast. It announces itself as a subcontractor who bid four crews and can staff two. Your durations were built on assumed crew sizes and unit rates; when the crew is half the assumed size, the duration roughly doubles, and you find out in week three of a fourteen-week activity. Money you can forecast, price, and carry contingency against. Absent people you cannot buy at any price — you can only replan around them, and replanning consumes float you have already spent. This is why 20.8's crew arithmetic is a scheduling skill, not an HR topic, and why the retention list in 20.9 is a project-controls document.

Project Checkpoint: The Willow Street Labor Plan

Your notebook currently holds a subcontractor management plan (Ch 19) with prequalification criteria, scope gaps, and a coordination structure. That plan governs the roughly eighty-five percent of Willow Street you are buying. This chapter's deliverable governs the fifteen percent you are building yourself — and it is the only part of the job where you own the productivity outright.

Build a Labor Plan for the Willow Street Community Center. Source data is Appendix K; forms are Appendix D. Six parts.

Part 1 — Crew compositions for the self-performed work. Kestrel self-performs concrete, rough carpentry, and general trades. On Willow Street that means foundations and slab, the wood-framed second floor and roof, blocking and backing, door and hardware installation, specialties, and final clean. For each, write the crew: how many of each classification, who the foreman is, and — the part people skip — the optimum size and why. Note the work-face ceiling for each operation, because the gymnasium framing and a corridor blocking run do not hold the same number of people.

Part 2 — Man-hour budget by cost code. A table: cost code, description, quantity, unit of measure, budgeted unit rate in MH per unit, and budget man-hours. It will look like this, and your numbers should total in this neighborhood:

Then price it at a blended burdened prevailing-wage rate — Willow Street is a municipal job — and state the rate you used and where it came from.

Part 3 — Weekly productivity targets. For each cost code, the target unit rate in man-hours per unit and the quantity that must be installed each week to hold the schedule. This is the sheet the foreman gets, and it must be in units a foreman can count.

Part 4 — Manpower loading curve by trade. Week-by-week planned headcount for every trade on the job, self-perform and subcontracted, summing to a total site curve. Mark the peak, and check it against the site: parking stalls, sanitary units, break area, and the gate. If your peak needs 62 people and you planned 40 parking spaces, you have found something.

Part 5 — Certified payroll process. One page: who pulls the wage determination and when, how classifications get assigned and by whom, how a worker performing two classifications in a day gets split, who signs the statement of compliance, the submission deadline and method, how apprentice ratios are checked weekly, and what you require from subcontractors and their lower tiers. Include the sentence "verified [determination source] on [date]; re-verify before award."

Part 6 — Your three leading indicators. Name the three things you will watch weekly to know productivity is slipping before the cost report tells you. Defend your three. Mine would be: (1) the weekly productivity factor trend by cost code, where three consecutive weeks below 0.95 triggers a walk-through, not an email; (2) overtime as a percentage of total hours, where any week above ten percent needs a written reason; (3) crew-mix drift — actual composition on the daily report against the composition in your Part 1 plan, because a crew that quietly lost two journeymen will produce a bad number four weeks from now. And one meta-indicator worth more than all three: the daily quantity report completion rate. If foremen stop turning in quantities, you have already lost the ability to measure anything.

Next chapter puts iron under the people: in Chapter 21 you will build the Willow Street equipment plan — own versus rent, crane selection, and an equipment schedule with costs. Your crew sizes from Part 1 are an input to it, because a crew waiting on a lift is a crew you are paying for twice.

Chapter Summary

A reference card, not a recap.

The five formulas

Reading a productivity factor

The diagnostic order when a labor code goes bad

1. Is the measurement real? Recount the quantity. Pull the time cards. Wrong inputs make every downstream conclusion wrong.
2. Did the work environment change? Access, sequence, trade stacking, material at the face, information.
3. Did the crew change? Size, mix, foreman, overtime, absenteeism, turnover.
4. Is there rework hiding in the code? Check the nonconformance and RFI logs.
5. Only then, and rarely, is it about effort.

The overtime arithmetic that people get wrong

• An overtime hour costs the burdened rate plus the loaded half-time premium on base wage — about 1.38×, not 1.5× the burdened rate.
• Sustained 10×6 for four weeks buys marginal production at roughly 1.8× straight time overall and about 2.5× by week four.
• Compressing a week (three 12-hour days) carries no weekly premium; extending one does. Check state daily-overtime rules and your labor agreement.
• Buy sequence, access, and crew size before you buy hours. Every time.

The burdened-rate checklist — base wage · paid time off · health and welfare · retirement · FICA · FUTA/SUTA · workers' compensation by class code × experience mod · general liability on payroll · small tools and consumables · overhead allocation. Multiplier typically 1.5–1.8, higher for low-wage classifications, higher again for high-hazard trades.

The prevailing-wage checklist — current wage determination in the file with a verification date · classification assigned by what the hands are doing · split hours for split classifications, recorded contemporaneously · fringe paid for all hours worked · overtime premium computed on the basic rate per the applicable act · apprentice registration verified and ratio checked weekly · certified payroll submitted on the required schedule with the statement of compliance signed · subcontractor payrolls collected and spot-audited against the site roster.

The three sentences to carry out of this chapter

1. Labor is the only major cost you control, and productivity is the term that moves.
2. Productivity is a management variable before it is a worker variable — sequence, access, hours, and crew design are decided in an office.
3. What you do not measure in man-hours per unit, you will explain in dollars per month, and by then it is a post-mortem.

What's Next

You now know what a crew costs and what makes it fast or slow. Chapter 21 puts machinery underneath them — own versus rent, crane selection and lift planning, hourly ownership and operating cost, and how to match equipment to the work so that the crew you just designed is not standing around waiting on a lift. The two chapters share a lesson: the most expensive thing on a construction site is a productive resource with nothing to do.