In the museum of work there is a corridor lined with doors that look identical until you try to pass through them. Some yield with a click. Some require a key that was issued in another state or another century. Some are painted with words like training, trade secrets, public safety, garden leave, reciprocity. A few are not doors at all but mirrors; you move toward them and meet your own fear of variance.
This essay is a cartography of those doors. It is not a pamphlet of outrage. We will measure the hardware, map the hinges, and note the locks that were added to keep out thieves but ended up keeping in apprentices. We will ask why a nurse licensed in one jurisdiction becomes an unlicensed stranger one exit down the interstate; why a junior engineer is told that the trick to learning is to stand still; why a barber can cut hair in one town but becomes a public hazard in the next. We will read the small print on the mirrors.
This is not a morality play, though morals intrude. It is a study of friction—the invisible tax on movement—and of its price in wages, new firms, and the quiet dignity of choosing one's next chapter.
I. The House of Locks
Two families of impediments govern the corridor. The first is contractual: noncompete agreements that declare a departing worker shall not ply their craft within a certain industry, radius, or period. The second is statutory: occupational licensing regimes that grant the right to practice a trade in exchange for training hours, examinations, fees, and, often, duplication for the sin of crossing a state line.
The first is private law rubber-stamped by courts; the second is public law enforced by boards. They are cousins. Each has a story in which it is necessary. Each has a shadow in which it is excessive. Together they define the mobility frontier of a labor market. Together they create the peculiar American condition in which a person can vote for president from any county but cannot cut hair, wire a thermostat, or change hospitals without seeking a new permission to be themselves.
The corridor's most recent renovation began with a federal experiment: an agency's attempt to abolish most noncompetes at a stroke. The result, after a season of litigation, is a cautionary footnote: as of September 2025 the federal noncompete ban is vacated; the door that seemed to open onto a national commons now opens onto case-by-case enforcement and state mosaics. The mirror remains, though the reflection changed costumes.
II. An Inventory of Friction
To reason about mobility as engineers rather than partisans, give the corridor a coordinate system. Define a Friction Index for any occupation o in any metro m:
$$\mathcal{F}(o,m) = \alpha \cdot \mathsf{NC}(o,m) \;+\; \beta \cdot \mathsf{LIC}(o,m) \;+\; \gamma \cdot \mathsf{PORT}(o,m) \;+\; \delta \cdot \mathsf{RISK}(o,m) \;+\; \epsilon \cdot \mathsf{COST}(o,m)$$
- NC measures contractual barriers: prevalence, scope, and expected enforceability of noncompetes (including chilling effects where workers believe clauses are enforceable even when courts would scoff).
- LIC measures licensing barriers: whether a credential is required; whether reciprocity or "universal recognition" exists; whether exams or supervised hours must be duplicated.
- PORT measures portability of benefits and training: Does health insurance travel? Do pension credits and certifications? Are apprenticeship hours recognized across borders?
- RISK measures consequences of violation: injunctions, damages, board discipline.
- COST tallies the monetary and time costs to comply, relative to wages in that occupation.
Normalize the scale so that $\mathcal{F}=0$ is a frictionless fantasy and $\mathcal{F}=100$ is a professional cul-de-sac. The coefficients are not philosophy; they are parameters estimated from data on job changes, wage growth after moves, and firm entry in linked employer-household panels.
What matters is not the elegance of the formula but its uses. A planner can overlay $\mathcal{F}$ on a map and observe where wages are low because outside options are weak; an economic-development office can tune incentives toward de-friction rather than the myopic theater of ribbon cuttings.
III. The Arithmetic of Outside Options
Wages are negotiations against the shadow of where else you might go. Raise the cost of moving and you lower the price of staying. This is not ideology; it is algebra. When employer concentration and legal frictions conspire, wages fall—not because employers are cruel but because the choice set has been compressed. Evidence from quasi-experimental work on employer concentration shows it depresses pay; legal frictions make that concentration bite by preventing exit at the moment it would help the most.
If you prefer a story: imagine two machinists in adjacent metros. The first can change shops tomorrow; the second must wait a year due to a covenant and must sit for an exam to use a lathe she has used for a decade. The two machinists are not the same person for wage-setting purposes; the latter negotiates with a shadow that has been made smaller by law.
IV. The Mathematics of Fear
One reason noncompetes persist is not the defense of the rare algorithm or the secret recipe; it is variance aversion. Managers prefer the stillness of a team that cannot be poached to the turbulence of a market in which talent moves. Legislatures tell themselves that duplication of exams protects the public. Boards convince themselves that a one-year restraint is a gentleman's handshake. The arithmetic is tidy: less variance, fewer surprises.
Yet friction is not a free lunch. It buys stillness by taxing learning by moving—the itinerant education of seeing how a craft is practiced across two floors, then across two firms, then across two regions. The literature calls this "knowledge diffusion"; workers call it getting better.
V. The Mirrors That Enforce Themselves
A perversity: even where a noncompete would likely be tossed in court, workers behave as if it were enforceable. Surveys suggest roughly one in five workers is bound by a noncompete at any given time and many more have signed one in their careers; the belief in enforceability—nurtured by HR scripts and exit-interview rituals—does much of the work. In other words, the corridor sometimes locks itself.
Policymaking that dwells on formal enforceability while ignoring this psychological lock-in designs for the wrong labyrinth. Nudges matter: mandatory plain-language disclosures that a clause is void in a given state lower $\mathcal{F}$ even if the statute was unchanged.
VI. The Country of Licenses
Licensing's defenders tell a noble story: charlatans are kept at bay, standards are kept high, the public is not harmed by impostors. Some of this is true; there are domains—anesthesiology, structural engineering—where the cost of error is a human life or a bridge. Much of licensing, however, is the fossil of a time when guilds chose to draw their borders on political maps.
Consider the scale. In the most recent labor-force data, roughly a quarter of U.S. workers report holding a government license; among some professions, the share is far higher. This is no longer a marginal institution; it is a dominant climate.
Licensing does many things. It raises wages for insiders (sometimes for good reasons), lowers geographic mobility, and creates deep nonlinearities: a worker 98% qualified somewhere becomes 0% qualified across a river. When a state adopts universal recognition—accepting out-of-state credentials rather than forcing repetition—it lowers $\mathcal{F}$ not by deregulating the craft but by deregulating the duplication of the same training. Arizona's 2019 law became the canonical example; dozens followed in some form.
If you are an engineer of systems, this is the lesson: much of the harm comes not from the existence of standards but from their refusal to travel.
VII. The Law's Attempt at a Shortcut—and Its Reversal
The federal agency's plan was elegant in its simplicity: declare most noncompetes an "unfair method of competition" and sweep them from the corridor, leaving trade-secret and non-solicit enforcement to do the work of protecting legitimate interests. The opposition was equally simple: the agency lacked authority to legislate by edict; the rule was overbroad; courts would balk. Courts balked. The ban was blocked in August 2024 and, after appeals and procedural maneuvers, the agency acceded to vacatur in September 2025. The national lever became a drawer of local tools again.
One can treat this as defeat, or as instruction. The corridor will not be bulldozed from above; it will be redesigned, room by room.
VIII. The City as a Machine for Learning (and the Wage That Follows)
Cities reward optionality. A programmer learns a new stack by walking two blocks; a nurse learns a new device by switching hospitals across town; a barber learns a new clientele by moving three stops down the line. If the corridor's doors are unlocked, the city is a machine for compounding tacit knowledge.
Economically, the payoff is captured by two curves:
- The Mobility Wage Curve: expected wage growth as a function of the number of distinct employers within commuting distance, adjusted for the cost of switching. It is steeper where $\mathcal{F}$ is lower.
- The Entry Elasticity Curve: new firm registrations per year as a function of the expected ease with which early hires can be assembled. Where noncompetes and redundant licenses abound, the elasticity flattens; entrepreneurs spend their nascent years drafting agreements rather than building teams.
The point is not that every switch is noble or every startup wise. It is that option value is a public good: your ability to credibly threaten to leave raises my wage too, because it forces the employer to compete on more than loyalty.
IX. The Mobility Atlas: A Work Plan
Treat this essay as a commissioning memo. Build a Mobility Atlas for three metropolitan regions—say, a logistics hub, a hospital-dense coastal city, and a manufacturing corridor. The deliverables:
Friction Maps by occupation: color tiles at the ZIP-code level with $\mathcal{F}$. Inputs: statutory texts; board rules; court outcomes; employer policies; survey data on perceived enforceability; typical contract carve-outs.
Reciprocity Ledger: for each license, list the states that recognize it, the conditions, and the cycle time to port.
Switching Cost Calculator: "If I am a nurse with X years and license Y in jurisdiction A, what is the expected time/cost to work in jurisdiction B?" Model includes exam duplication, jurisprudence add-ons, CE credit conversion, application queue times.
Outside Option Index: for each census tract, compute the number of employers within 45 minutes who can legally hire an incumbent in occupation o within 60 days, given the frictions. This is what the negotiator should care about.
Before/After Scenarios: simulate the effect on wages and vacancies of (a) voiding noncompetes for workers under pay threshold T; (b) universal recognition of licenses; (c) portable benefits and training credit transfer.
The methodology is not speculative. It is empirical labor economics plus administrative law, sharpened into pictures that even a hurried mayor can understand.
X. Case Vignettes (Without Romance)
The Itinerant Nurse. She has 12 years' experience in acute care. A neighboring state pays 9% more and offers a schedule that would let her care for an aging parent on Thursdays. Without universal recognition, the move requires an exam, a jurisprudence module, an application fee, and a queue measured in months. With recognition, the module remains; the duplication disappears. The hospital in the receiving state fills a vacancy in three weeks. The patient touched by this timeline never learns why the nurse arrived sooner; the wage learned it.
The Apprentice Electrician. He has accumulated 1,900 supervised hours; the neighboring jurisdiction requires 2,000 with a different mix. Without reciprocity, the move resets the counter as if hours worked elsewhere have no moral weight. With reciprocity, the last 100 hours are counted; inspection goes forward. Thirty years from now the electrician will describe his craft to a grandchild and the story will include two methods learned from two masters; the building he wired will still be standing.
The Junior Engineer. She signed a document on her first day that she barely read. She now knows that the clause might be void where she lives but believes—because someone in HR read a script—that it will be used to "make it painful." She stays, though a colleague moves and doubles his learning rate. Five years later she is still good at one thing. She is loyal; the wage is not.
XI. Reforms That Respect Secrets Without Selling Silence
It is possible to protect legitimate interests without turning the corridor into a cul-de-sac. A reformer's toolkit:
Bright-line noncompete bans for workers below a generous pay threshold, paired with robust trade-secret enforcement (DTSA/UTSA) and non-solicit clauses that are narrow and time-boxed. The defense of what is truly proprietary should be surgical, not totalizing.
Universal recognition for licenses in fields where safety is real but duplication is theater. Keep the jurisprudence module; delete the exam retread. Reward boards that publish cycle-time dashboards so delays can be audited like hospital wait lists. Arizona's statute is a useful template precisely because it honors standards while abolishing the ritual of repetition.
Portable benefits and training credits. Let employer-funded training vest like a pension: the worker who leaves early pays back a declining balance rather than facing a punitive clawback. Recognize apprenticeship hours across borders the way universities transfer credits—with published equivalence tables and a presumption of good faith.
Disclosure mandates: where noncompetes are void or limited, require employers to say so in 14-point type at hire and at exit. Replace the fear script with the truth. (A surprising portion of the corridor's force derives from ignorance.)
Targeted garden leave for genuinely sensitive roles, paid and narrow. If a firm fears opportunistic poaching in the wake of product launches, pay for a short intermission; do not exile the chorus.
Public procurement preferences: favor vendors that forswear noncompetes for rank-and-file and participate in reciprocity compacts. Governments already buy; they can also teach with their purchases.
XII. The Economics of Training: Who Pays for the Ladder?
A common defense of noncompetes is that firms will underinvest in training if workers can depart the moment they are useful. This is a version of the appropriability problem: returns to investment diffuse to rivals. The answer is not a blanket ban on exit but contracting for the ladder with specificity:
Define training that creates general human capital (widely valuable) and training that creates firm-specific capital (valuable mostly here). Let the former vest over time and travel with the worker; let the latter be protected by targeted repayment if departure is immediate and opportunistic.
Establish industry training funds co-financed by clusters of firms and local government. The cost of general upskilling is socialized among beneficiaries; no firm needs to bolt the door to guard against paying for a competitor's future.
Use wage insurance pilots to cushion short-run losses when workers switch into adjacent fields. The corridor is less frightening when the floor is visible.
XIII. The Hospital Ward, the Barber Chair, the Datacenter
Licensing, at its best, protects patients. At its worst, it becomes a map of turf. Consider three archetypes:
Medicine and nursing. High stakes are real. Reciprocity should be the norm; duplication should be reserved for genuine differences in scope of practice. A national compact with published portability checklists changes lives without changing standards. Delay is not caution; delay is harm.
Personal services. Public-health rationales exist; so does guild rent-seeking. Evidence-based hours, reciprocity, and frequent sunset reviews prune back growths that once were justified and now are simply there.
Digital professions. Here licensing is rare (for now) but noncompetes are prevalent in mid-tier roles. Replace them with non-solicits and IP assignment clarity. Where sensitive architectures are concerned, use paid cooling-off windows, not open-ended exile.
XIV. The Structural Counterpart: Concentration Without Exit
Even if you never sign a covenant and your board grants you a license in gold leaf, your wage will sag in a market where three employers control the relevant job tier and the fourth is two hours away. Mobility reforms are therefore complements, not substitutes, for competition policy. When fewer firms exist, every unit of friction bites harder. When more firms exist, the corridor is less a gauntlet and more a boulevard.
Empirical work on employer concentration reminds us that the wage question is not solely about worker power; it is also about outside options engineered by market structure. The most elegant noncompete reform will disappoint if the town has only two large buyers of your skill.
XV. A Note on Measurement (So We Don't Lie to Ourselves)
To keep this essay from becoming a sermon disguised as a map, commit to measurement:
Prevalence: update estimates for how many workers are bound by noncompetes, by wage band and occupation. Treat the GAO's and survey literature's 18% figure as a baseline and re-measure after reforms.
Licensing share: track the proportion of workers licensed each year and by region; use the household survey tables as the denominator, not anecdotes. If the share remains at roughly one quarter nationally, demand justification where it climbs.
Cycle time: for each licensing board, publish application processing times and reciprocity approvals like airport on-time dashboards.
Exit elasticity: compute how likely a worker is to switch employers after a negative shock. If reforms work, the elasticity should rise and wages should rise without increased separation into unemployment.
Entry: count new firms in friction-sensitive sectors before and after reciprocity adoption. If the corridor opens, the city learns faster.
XVI. Objections, Answered with Respect
"We will lose secrets." Secrets deserve defense. The law of trade secrets is a toolkit that includes injunctions, damages, and, in egregious cases, criminal law. Use it. Keep the defense proportionate and targeted, not a blanket encumbrance on everyday movement.
"We trained them." Then share the ladder; let the worker repay a declining portion if departure is immediate and the training is unusually firm-specific. Make the repayment schedule public to deter abuse.
"Reciprocity lowers standards." Reciprocity lowers duplication. Keep jurisprudence modules to teach local law. Audit outcomes. If quality falls, adjust. If it doesn't, retire the myth that borders make the barber's scissors safer.
"Litigation risk will explode." It falls when contracts are clear and narrow. Ambiguity breeds lawsuits; precision prevents them.
"The federal ban failed, therefore the project is doomed." The failure teaches federal humility and state creativity. Corridors are local; maps can be redrawn without waiting for an edict.
XVII. The Architect's Detail: A City Contract Clause
If you run a city that buys services, add a page to every RFP:
- Vendors certify they do not use noncompetes for workers below a specified pay threshold.
- Vendors commit to participate in a reciprocity compact for licensed roles.
- Vendors publish training vesting schedules with declining repayment.
You are not legislating; you are buying. The corridor will change because your money walks before the law crawls.
XVIII. The Calendar
A generous schedule:
Within 90 days: adopt disclosure mandates; publish licensing dashboards; begin reciprocity negotiations with neighboring states; instruct procurement to add the clause.
Within 6 months: pass threshold bans for noncompetes; stand up a training-credit portability registry; sign reciprocity compacts for at least five high-vacancy roles (nursing, respiratory therapy, electricians, plumbers, early-childhood educators).
Within 12 months: publish the first edition of the Mobility Atlas; refresh annually; adjust coefficients as new data arrives.
No miracles, just carpentry.
XIX. The Country Ahead
Imagine a corridor with fewer mirrors and more doors that open the same way regardless of which county built them. The nurse moves one exit down the interstate to care for a parent and does not spend six weeks awaiting permission to be useful. The barber learns a new clientele and brings the style back home. The apprentice electrician finishes his hours without pretending they never happened. The junior engineer signs a narrow non-solicit and a robust trade-secret clause and learns two architectures in five years, not one in ten.
Wages respond, not out of generosity, but because the shadow of choice lengthens. New firms are formed because assembling early hires is no longer an impossibility proof. Training increases because it is priced correctly and made portable rather than hoarded as a pretext for captivity. Boards discover that variance was fear's word for learning.
The corridor will never be perfectly smooth. Nor should it be; standards are guardrails; pauses can be prudence. But a civilization that confuses friction with virtue becomes a museum of stalled apprentices. The remedy is not the sledgehammer of slogans but the file and oil of design.
XX. Coda: The Map and the Mirror
A poet once suggested that maps and mirrors are both instruments of knowledge and illusion. The noncompete and the license have been both: they have protected, and they have also convinced us that stillness is safety. If we must keep one mirror, let it reflect the trade secret and nothing else. If we must keep one map, let it be the Mobility Atlas—a living chart in which each corridor is scored, each door labeled, each hinge audited, and each lock justified anew.
The price of movement is the price of becoming. When a society lowers that price, it is not merely raising wages. It is enlarging the number of plausible futures a person can credibly say aloud.
Notes & References
Key legal timeline: the federal noncompete rule announced in April 2024 was blocked by a federal court in August 2024; in September 2025 the agency moved to dismiss its appeals and acceded to vacatur—the rule is not in effect.
Prevalence baseline: nationally representative surveys and GAO synthesis estimate ~18% of workers are bound by noncompetes at a point in time, with a far larger share reporting they have signed one at some point.
Licensing scale: about a quarter of the U.S. workforce held a government license in 2024.
Reciprocity template: Arizona's 2019 universal license recognition law as the canonical model for cross-state portability.
Outside-option economics: increases in employer concentration causally depress wages; frictions magnify this effect by making exit costly.
Hemispheric comparisons:
- Spain: Workers' Statute Art. 21 requires compensation for post-employment noncompetes; durational caps apply (typically ≤2 years technical roles, ≤6 months otherwise).
- Mexico: Constitutional right to work renders most post-employment noncompetes unenforceable.
- Brazil: Enforces narrow noncompetes only with substantial compensation (often ~1 month salary per month of restriction).
- Colombia: Permits tightly tailored covenants with safeguards.
State reforms: Minnesota (2023 near-total ban); California SB 699 & AB 1076 (2024 expansions); Military Spouse Licensing Relief Act (2023 federal portability mandate).
Academic evidence: Marx, Starr, Johnson, Kleiner, and others on noncompete enforcement effects on wages, entrepreneurship, and geographic mobility.
(Figures and legal statuses reflect developments current to November 7, 2025.)