Section 1
The reflex that breaks
The standard move when you cannot hire is to reach for the wage lever. It is the first thing every business owner learns and the first thing every consultant recommends. It works beautifully in a normal labor market, because in a normal market, supply slopes upward. Offer more, and more people show up. Offer a lot more, and you pull workers from neighboring firms, from adjacent trades, from retirement, from the next town. The reflex breaks when supply stops sloping upward. Economists call this inelastic supply, which is a dry phrase for a brutal reality: the quantity of qualified people does not change much no matter how high the price goes, at least not on any timeline that helps you this quarter or this year. You cannot conjure a licensed master electrician with fifteen years of field time by raising your rate twenty percent. The training pipeline for that person is a decade long, and it was underfilled a decade ago. The evidence that we have crossed into inelastic territory is no longer anecdotal. In Germany, the Confederation of Skilled Crafts, the ZDH, reports more than twenty thousand unfilled positions in the craft trades even as roughly 350,000 apprentices move through the system, and the German Federal Employment Agency counted around 225,000 unfilled skilled positions for vocationally trained technicians and craftspeople in 2025, most concentrated in the building trades, with construction electrics at the top of the list. Those are not gaps that a signing bonus closes. They are structural. And the labor market is not even cooperating on the applications you do get. Indeed's ghosting research found that roughly 89 percent of employers now report job seekers dropping out of the process or failing to show on day one as a real problem, a number that has climbed as the market tightened. (One honest flag: that 89 percent is an all-sector figure, not a trades-specific measurement, so read it as a directional signal about hiring friction rather than a precise trades statistic.) The point stands either way. Even the top of the funnel leaks before you get to talk about money.
Section 2
Model one: comparative statics, or why price does not equal supply
To see the ceiling clearly, start with the plainest formal tool in the box: comparative statics. You change one variable, hold everything else still, and trace where the equilibrium moves. In a healthy market for technicians, push the wage up and the quantity supplied rises along the curve. More people enter, poaching gets easier, and your capacity grows roughly in step with what you are willing to pay. The lever and the outcome are connected. Now make the supply curve vertical, which is what inelasticity means in the limit. Push the wage up and nothing happens to quantity. The line does not move because there is no line to move along. You have paid more for the same number of people, or more precisely, you have paid more to take a person away from a competitor while the total pool stays fixed. Your cost went up. Your capacity did not. This is the core reframe in one sentence: when supply is inelastic to price, spending more converts a cost problem into a redistribution fight, and redistribution fights do not add capacity to the market. They just decide who holds the fixed capacity that already exists. Once you accept the vertical curve, three things about your own business change meaning. First, growth stops meaning what it used to. In the old model, growth was a demand question. Win more work and grow. Under a capacity ceiling, demand is not your constraint, so winning more work does not grow you. It only lengthens your backlog and raises the odds you disappoint someone. Growth now means one of exactly three things: getting more output from the people you have, changing the mix of work toward higher value per hour, or genuinely expanding the pool through training you fund yourself. Anything else is motion, not growth. Second, a good year stops meaning a full calendar. A booked-solid schedule used to be the trophy. Under the ceiling, a booked-solid schedule at your old rate is a signal that you underpriced your scarcest asset, which is your own delivered hours. A good year is now a year where revenue per available technician-hour went up, not a year where the calendar was full. Those are different targets, and chasing the second one keeps you poor and exhausted. Third, turning away work becomes a strategy rather than a failure. If you cannot add capacity, then every job you accept is a job you chose over another job. The operators who thrive under a ceiling get deliberate about which work they decline. The ones who struggle say yes to everything, then subcontract the overflow to someone slower or shakier and inherit the callbacks. Comparative statics (price does not equal supply), the first-order lens • Assumes: you can hold everything else still and move wage alone; the supply curve has a knowable shape. • Fits because: the trades shortage is fundamentally a question of which way one lever pushes the equilibrium. • Breaks when: the supply curve is not actually vertical, only steep. Some slack always exists at some price (retirees, adjacent trades, migration), so treating the ceiling as perfectly fixed makes you give up on real, if expensive, capacity. • Counteracts: the reflex that any hiring problem yields to a bigger number. • May reinforce: fatalism. The model can talk you out of recruiting entirely when smart recruiting still moves the margin.
Section 3
Model two: game theory, or why the ceiling can drop on you
Comparative statics tells you the pool is fixed. Game theory tells you that a fixed pool is exactly the condition under which other players start fighting you for your share of it, and one class of player has changed the rules. Set the game up simply. The players are the independent operators in your market plus, increasingly, private equity backed roll-ups buying and consolidating trades businesses. The resource is the fixed pool of qualified technicians. The moves are how much each player is willing to pay, in cash and in non-cash terms, to hold or take a technician. When every player is a similar independent shop, the game reaches a rough equilibrium. Nobody can pay wildly more than the work supports, so poaching happens at the margins and the pool sorts itself by relationships and reputation. The ceiling sits where the economics of the work put it. A private equity backed consolidator changes the payoff structure in a way that should worry you, because it is not playing the same game. Its return does not come primarily from the profit on the jobs. It comes from buying businesses at one multiple of earnings and selling the combined entity at a higher multiple. Technicians are the input that makes the acquired businesses worth anything, so the consolidator can rationally pay more for that input than the job economics alone would justify. It is buying a person to protect an acquisition thesis, not to break even on next Tuesday's service call. That is technician poaching with a structurally deeper pocket, and it does something specific to you: it lowers your effective ceiling. Not the market's total pool, which is unchanged, but the slice you can realistically hold. Every technician a consolidator can pay above job-economics to take is a technician removed from your reachable supply. Your vertical line just shifted left. You now sit under a lower ceiling than you did before the capital arrived, and you did nothing wrong to earn it. This is the part operators miss because it is invisible on the day it happens. You do not get a memo that the ceiling dropped. You just notice, six months later, that the tech you would have hired went somewhere that paid more than the work should support, and you assume you were outbid on price. You were, but the price was set by an acquisition math you are not in a position to match, and no amount of you paying more fixes it, because you are still playing the job-economics game and they are not. Game theory (technician poaching), the strategic lens • Assumes: identifiable players choosing pay-and-retain moves over a shared, fixed resource. • Fits because: a fixed pool plus a new player with a different payoff function is the textbook setup for a resource-allocation game. • Breaks when: the consolidator's thesis collapses. Roll-ups can overpay into a rate environment that no longer supports the multiple, at which point the deep pocket empties and the ceiling springs back up. Do not assume the poacher is permanent. • Counteracts: the assumption that losing a hire always means you were simply too cheap. • May reinforce: paranoia. Not every competitor is a funded roll-up, and reading capital into every lost candidate wastes attention.
Section 4
Model three: the threshold, watching for the sudden tip
The first two models describe a steady state. The third watches for the moment it flips. Threshold models exist for phenomena that go nothing, nothing, nothing, then everything at once, and a capacity ceiling has one of those moments buried in it. The tip happens when demand, backlog, and your own physical limit collide. For a while you absorb rising demand by working longer, compressing your schedule, and stretching your best people. The system looks fine from outside. Then you cross a threshold where the backlog is long enough that customers stop waiting and go elsewhere, or your best technician, the one holding the whole thing up, burns out or gets poached, and the compensating slack you were quietly running on vanishes. Output does not decline gently. It steps down, because it was being held up by one or two people operating past sustainable load. The practical lesson is to stop watching the average and start watching the buffer. The relevant number is not your average utilization. It is how much reserve capacity sits between your current load and your true ceiling, and specifically how much of your output depends on a single person you cannot replace. When that buffer approaches zero, you are one resignation away from a step change, and the threshold does not warn you before it fires. Threshold, the tipping lens • Assumes: a reinforcing dynamic and a critical point where behavior flips discontinuously. • Fits because: capacity held up by a few overloaded people fails suddenly, not gradually. • Breaks when: the system is more resilient than it looks and degrades gracefully instead of tipping, which makes threshold thinking cry wolf. • Counteracts: the comfort of watching averages that hide single-person dependencies. • May reinforce: alarmism that treats every busy stretch as an imminent collapse.
Section 5
The structure-break flag
Every model above assumes the structure it runs on holds still. The honest move is to name the thing that could change the structure underneath all three at once. Here is the flag for this piece. The models assume the fixed pool and the funded poacher are the stable backdrop. The structure breaks if two forces meet: a demand stimulus that was propping up the whole picture reverts at the same moment the absolute supply floor is reached. If a subsidy, a boom, or a wave of deferred maintenance was inflating demand, and it switches off while the pool of people is at its hard minimum, you get whiplash in both directions. Demand can fall faster than you expect while supply stays stuck, and the poacher's deep pocket, which depended on a rising rate environment, can empty overnight. The ceiling does not just move. The whole room changes shape. Watch for the reversion, because that is the day your model stops describing your market.
Section 6
The proposed solution: GEER, then RADAR, then CHAIN
Diagnosis without a lever is just anxiety with footnotes. Run three passes. GEER: rank the levers you actually control GEER means translating the situation into concrete, ordered levers, cheapest and most reversible first. Under a capacity ceiling, the dominant channels are output-per-person, work-mix, and your own pool-building, not the wage. So the lever order looks like this. Pull first, because they are cheap and reversible: raise price on your scarcest work until the calendar is no longer full at every rate, since a full calendar at your old rate is proof you were subsidizing demand you cannot serve. Cut the low-value work off the bottom of the mix so your fixed hours land on higher-margin jobs. Remove the non-technical tasks from your best technician's day, because every hour of paperwork you take off the ceiling-setting person is the cheapest capacity you will ever buy. Pull second, because they cost more but still sit in your control: systematize the work so a lower-tier hire can do the routine eighty percent under supervision, which turns the scarce master's time into oversight rather than execution. Invest in the tools and software that shave real minutes per job, since minutes multiplied by a fixed headcount is the only free capacity on the table. Pull last, and flag it honestly as the slow, expensive lever: build your own pipeline through apprenticeship you fund and train, accepting that it pays off in years, not quarters, and that some of the people you train will be poached before you recoup them. It is still the only lever that adds capacity to the market rather than redistributing it, which means it is the only lever that lifts the ceiling instead of fighting over who sits under it. Note the no-lever flag: there is no cheap, fast lever that adds a licensed master this year. If your plan requires one, your plan requires financing a multi-year training bet or accepting a lower growth ceiling. Pretending otherwise is the mistake. RADAR: a dated portfolio under real uncertainty You do not know whether the poacher stays or the stimulus reverts, so do not pretend to. Sort your actions by reversibility instead. Do now, because they are near zero-regret in every scenario: reprice the scarce work, strip non-technical load off your key people, and start measuring revenue per available technician-hour as your real scoreboard. These help whether the market tightens or loosens. Hedge, as cheap insurance against the single-person tip: cross-train a second person on the one skill that currently lives in one head, and document the work so a departure is a setback rather than a collapse. It costs a little now and caps a large tail loss. Defer, but pre-commit the trigger: hold the decision to fund a full apprenticeship pipeline or to sell into a roll-up until you see a specific observable. Trigger on it in advance. For example: if you turn away more than a set share of qualified demand for two consecutive quarters and your buffer to ceiling stays near zero, that is your signal to commit to the multi-year training bet, because the ceiling has proven binding rather than temporary. CHAIN: check yourself against the base rate Before you believe your own story, match it to history on structure, not surface. The reference class here is any industry that hit an inelastic input constraint while capital moved in: think trucking during driver shortages, or nursing, or any trade that consolidators have entered before. The base rate from those cases is sobering and useful. Wages rise, margins compress for holdouts, a wave of consolidation buys up the operators who cannot solve the capacity problem, and the ones who survive independently are disproportionately those who moved up-market, controlled their mix, and built their own training rather than bidding on a fixed pool. Tilt that base rate by your present state: how binding is your ceiling really, how funded is the capital in your specific market, how close is your buffer to zero. Then subtract the counterfactual. Some of what feels like a capacity crisis is just a hot cycle that will cool on its own, and the matrix-break flag applies. If the stimulus reverts, the base rate from boom conditions stops predicting your next two years.
Section 7
The blind spot
Here is what this ensemble cannot see, stated so you do not mistake the framework for the territory. All three models treat the definition of the work as fixed. They assume the job still requires the person it has always required. They are blind to the possibility that the work itself gets redefined, by tooling, by prefabrication, by remote diagnosis, by software that lets a lower-skilled person deliver what used to need a master, so that the scarce input stops being scarce because it stops being needed in the same quantity. If that happens, the vertical supply curve does not move. It becomes irrelevant, because the demand for that specific skill drops out from under it. The models tell you how to survive a fixed constraint. They cannot tell you when the constraint quietly stops being the binding one.
Section 8
The fitness test
Run this on your own business before you touch a wage. Take your single most skilled person, the one whose calendar sets your real ceiling. Ask: if that person doubled their pay tomorrow and stayed, could you serve more customers next month? If the honest answer is no, you do not have a labor cost problem that money solves. You have a capacity ceiling, and your job is not to win a bidding war for a fixed pool. Your job is to get more output from that person, move your mix above them, and slowly, expensively, build the next one. Measure yourself in revenue per available technician-hour, watch your buffer to the ceiling, and treat the day a funded buyer enters your market as the day the ceiling dropped, not the day you got too cheap.