Albert Hirschman didn’t write Exit, Voice, and Loyalty to explain pharmacy benefit managers.

He wrote it to explain a more general puzzle: why institutions can decay for a long time in full view of the people who depend on them—and still survive.

One of his most vivid examples comes from Nigeria. Hirschman observed a railway system losing freight customers to trucking. On paper, that looks like the beginning of an existential crisis. Customers leaving. Revenue falling. Competition winning. The kind of pressure that’s supposed to force competence.

But the railway didn’t respond like an organization facing extinction. It didn’t suddenly become customer-obsessed. It didn’t treat the departures as a five-alarm fire.

Because the freight customers who left weren’t the ones keeping the railway alive.

The railway still had captive passenger traffic. People without alternatives. People who were too constrained, too resigned, or too indifferent to mount an effective complaint. The departure of the most demanding freight customers didn’t sharpen management’s attention—it dulled it. Exit didn’t deliver a clarifying signal. It removed the people who would have delivered voice.

And the truly dark part is this: the railway could welcome the departure of “high-maintenance” customers. Not because it was improving, but because it was quieting. Less freight meant less complaining, without threatening the system’s ability to keep collecting fares from those who couldn’t leave.

That story reverses a comforting assumption: that customers leaving is the ultimate accountability mechanism.

Hirschman shows how exit can sometimes protect an institution from accountability by filtering out the very people most capable of demanding improvement.

If you’ve spent any time around the PBM market, you’ve seen this dynamic—just with different names on the trucks.

Exit, voice, loyalty—and the missing fourth behavior

Hirschman’s core framework is deceptively simple. When the quality of an organization declines, people respond in three main ways.

They can exit—leave for an alternative.

They can use voice—complain, organize, demand changes, threaten to leave if quality doesn’t improve.

Or they can remain loyal—stay invested enough to tolerate decline and have faith in a turnaround, sometimes because loyalty is a byproduct of being an inert consumer.

Later researchers extended this into EVLN by adding a fourth response: neglect. The passive, corrosive decision to stop engaging. Stop checking. Stop expecting improvement. Stay, but quietly detach.

In a healthy competitive market, exit disciplines producers. Bad actors bleed revenue and either improve or die.

In a monopoly, exit is impossible, which can sometimes force voice. If people can’t leave, they have to complain. The complaints accumulate in one place. The signal lands somewhere.

PBMs sit in a more perverse zone.

Exit is technically possible, but functionally unhelpful.

Employers stay with PBMs they don’t trust because they’re living in quasi-lock-in—a condition where switching is feasible, but expensive and burdensome, and additionally rarely changes the underlying dysfunction. And quasi-lock-in can be worse than monopoly or competition, because it offers just enough escapability to bleed off pressure while preserving the structure that makes the system degrade.

Hirschman had a phrase for what follows when accountability mechanisms fail: “an oppression of the weak by the incompetent and exploitation of the poor by the lazy.” In the PBM world, “weak” doesn’t mean unsophisticated. It means structurally disadvantaged—smaller groups, fragmented purchasers, anyone forced to operate through opaque intermediaries with limited leverage and limited ability to verify what they’re being told. “Lazy” doesn’t mean unprofitable. It means extraction that doesn’t require brilliance because it’s embedded in the system’s design.

Not a heist.

A gravity field.

The PBM market as a railway with a trucking lane

Start with the surface-level fact everyone knows: large, sophisticated employers can “exit.”

They can run RFP scenarios. They can threaten to move from CVS Caremark to Express Scripts to OptumRx and back again. They can hire a consultant, benchmark guarantees, renegotiate terms, execute a transition, and declare victory.

But the PBM rarely experiences that exit as an existential event.

The PBM’s revenue base is diversified across buyer segments and lines of business. Losing a large employer matters, but it’s not necessarily fatal—especially in a world where the same corporate family may still touch those lives through other channels due to vertical integration. More importantly, even when an employer truly leaves, the PBM does not necessarily lose the market structure that enables its profits.

Because the employer doesn’t exit the PBM system.

They usually just exit one PBM and enter another that is structurally similar. Different logo. Different account team. Different set of contract exhibits. Same core features: opacity, multi-layered incentives, information asymmetry, and a pricing architecture designed to be negotiated rather than understood.

That’s why PBM churn is so often misread. We treat churn like proof the market is functioning—look, employers can switch; nobody’s trapped.

In practice, churn behaves like filtration.

It removes the most alert, most demanding buyers from a given PBM’s book of business and leaves behind the captive or the resigned—without forcing the model itself to confront accumulating pressure for change.

The railway lost the truckers. It kept the passengers.

PBMs may lose some high-scrutiny employers. They keep a vast population of smaller groups, MEWAs, public entities, and purchasers who either can’t mount an effective challenge or don’t have the tools to distinguish “better contract language” from “better outcomes.”

Exit becomes a safety valve.

It releases energy that might otherwise become voice.

Quasi-lock-in is not captivity. It’s worse.

Quasi-lock-in is not the same as being unable to leave. Employers can leave. They do leave.

The point is that leaving often fails to accomplish what exit is supposed to accomplish: a sharp corrective signal that forces improvement.

In quasi-lock-in, exit doesn’t deliver relief because the alternatives are equivalently dysfunctional.

Employers know this intuitively, even if they don’t have language for it. You can imagine it in the way benefits teams might talk in hallways, not in procurement decks.

“We’re thinking about moving, but to who?”

“We’ve been through two PBMs in six years and it’s the same movie.”

“We negotiated better guarantees, but we still can’t tell what we’re actually paying.”

“The rebate story changed, but the net trend didn’t.”

“We’re just moving dollars around.”

“Switching doesn’t help” isn’t cynicism. It’s a learned conclusion from repeated cycles inside a closed set of options.

What makes exit functionally unhelpful in PBMs isn’t a single barrier. It’s a stack. A layered architecture of plausible deniability.

The first layer is comparability. A PBM contract is less like buying a service and more like choosing a rulebook. And the rulebooks are engineered to resist clean comparison. Definitions move. Exclusions multiply. Baselines shift. Thresholds change. Specialty is whatever the contract says it is—until it isn’t because the value of going against the contract is more profitable than the level of penalty. You can “win” a guarantee while losing the game it’s supposed to represent because the guarantee is tethered to a category definition that drifts underneath you.

This is the oldest trick in this market: make the thing you’re promising flexible enough that it can always be technically true while practically meaningless.

The second layer is verification. Even sophisticated buyers struggle to validate whether promised economics are realized, because the truth lives in places employers don’t directly touch. Ingredient cost logic that turns on MAC lists you don’t get to interrogate in real time. Spread dynamics that show up as “discounts” and “effective rates” rather than a transparent pass-through of acquisition-linked pricing. Manufacturer rebate contracts you can’t see, paired with formulary decisions framed as clinical while functioning as economic levers. Utilization management criteria that change quietly, turning access friction into abandonment—and abandonment into savings that look like “trend control.”

The data exists. The problem is you don’t own the interface.

The third layer is switching costs. And not just financial. Switching imposes operational risk: member “disruption”, pharmacy network changes, prior authorization resets, specialty pharmacy operations transitions, patient confusion, internal communications burden. It’s a major project. Most employers do it under time pressure with limited internal bandwidth, leaning on intermediaries who themselves operate inside the same system.

Then comes the part Hirschman would recognize: the psychological reset.

After an exhausting switch, HR teams want peace. They want closure. They want to believe the new partner is different and quality is comparable. That desire doesn’t make them naïve. It makes them human.

But it also pushes the organization toward loyalty or neglect right when disciplined voice is most needed.

This is the paradox of quasi-lock-in:

Escapability enables durability.

Because you can leave, pressure never builds meaningfully.

Because leaving doesn’t help, nothing improves.

Because nothing improves, leaving becomes ritual rather than remedy.

Why quasi-lock-in can be worse than monopoly

It’s tempting to reduce this to concentration. “Isn’t this just a market with three dominant players?”

Hirschman’s insight is that concentration alone doesn’t determine behavior. What matters is how exit and voice interact.

In a monopoly, you can’t meaningfully exit, so the dissatisfied must either use voice or accept decline. That can be ugly, but it keeps the feedback loop intact. Complaints accumulate in one place. The institution can’t pretend the problem belongs to someone else.

In a competitive market with true differentiation, exit is powerful. Bad actors bleed revenue, and the remaining customers become a clearer signal. The market learns.

Quasi-lock-in breaks both loops.

Exit doesn’t discipline the provider because the provider isn’t meaningfully threatened by any single departure. And exit doesn’t strengthen voice because the people most capable of voicing dissatisfaction are exactly the ones most likely to leave.

The result is institutional stagnation with a glossy veneer of “choice.”

This is where Hirschman’s “lazy monopoly” becomes the right lens. PBMs are not lazy because individuals inside them don’t work hard. Many people inside PBMs work extremely hard. “Lazy” here means structural. A system that persists without earning trust because its rents come from complexity and position, not from excellence that buyers can clearly reward.

The “PBMs are geniuses” story gives them too much credit. It frames the whole thing as a brilliant plot. It turns structural advantage into mastermind strategy.

The more actionable truth is less cinematic and more damning: much of the extraction is ambient, not ambitious.

It looks like channel steering that quietly routes volume to owned assets.

It looks like “guarantees” that treat opacity as a feature, not a bug.

It looks like rebate-driven formulary behavior that keeps list prices inflated while buyers are trained to feel grateful for a pass-through check that arrives months later.

It looks like copay accumulator and maximizer logic whose practical effect is to move who pays, when, and whether the patient actually makes it to the first fill.

It looks like drift—subtle, continuous, easy-to-deny changes in access rules, billing practices, network design, specialty definitions, and reporting scopes. Nothing dramatic. Nothing that triggers an emergency meeting. Just a steady slide that only becomes obvious after the year is over and everyone is too tired to re-litigate it.

A lazy monopolist doesn’t need to win every negotiation.

It just needs to keep the system sufficiently illegible that purchasers can’t reliably tell whether they’re getting better or worse—and sufficiently interchangeable that “doing something” often means switching within the same architecture.

Churn as filtration, not failure

This is where the Nigeria railway story stops being metaphor and starts being diagnosis.

In the PBM market, churn is often interpreted as a retention problem: “We lost an account.”

But in quasi-lock-in, churn can be a feature. It removes the most vocal, high-scrutiny buyers and sends them elsewhere without changing the underlying economics of the model. The system stays intact. The signal is lost.

And the population that remains tends to be less likely to demand transparency, run deep audits, or invest in monitoring or surveillance—not because they don’t care, but because they’re constrained by scale, staffing, and expertise. Over time, that shapes the feedback PBMs receive. If the people most capable of detecting problems leave, the remaining customers provide less friction.

That creates a form of market selection that is the inverse of what competition is supposed to do.

Instead of rewarding excellence, the market rewards tolerability.

PBMs don’t have to be trusted.

They just have to be survivable under neglect.

Hirschman’s warning lands cleanly here: exit can be the enemy of reform. Not because leaving is wrong, but because leaving can drain the reform impulse from the very people with the leverage, language, and stamina to demand change.

Diversive energy: the treadmill that eats your bandwidth

Hirschman described “diversive” behavior—energy that feels like action but diverts attention from deeper change.

In the PBM world, the diversive energy is the endless cycle of lateral movement: RFPs, finalist presentations, contracting games, implementation war rooms, consultant fees, new ID cards, new specialty pathways, new exceptions processes, new reporting templates.

A tremendous mobilization of effort that often produces, at best, marginal improvements inside the same architecture.

This is not to say employers should never switch. Sometimes switching is rational. Sometimes it buys quality. Sometimes it fixes specific operational pain.

But zoom out.

If the only lever you can pull is “choose another Big 3,” then most of your energy goes to rearranging an opaque system rather than changing it.

That opportunity cost is the quiet tragedy of quasi-lock-in. The frontier stays underexplored because you’re stuck paying the tax of administration: building the next RFP instead of building the instrumentation that makes the problem legible.

The next era of pharmacy benefit strategy won’t be won by better negotiating scripts.

It will be won by better measurement infrastructure.

You cannot manage what you cannot see. And the PBM model has survived not only because it extracts value, but because it keeps the extraction difficult to observe and even harder to attribute to a specific mechanism.

Fog is not an accident in this market.

Fog is a strategy.

Coverage intelligence is the counter-move: my own Policy Vault, coverage crawlers, drift monitoring. Not as a side project. As an accountability layer. It turns the rulebook into a dataset.

Because what PBMs fear isn’t churn.

It’s receipts.

The rules can be found.

The rules can be compared.

The rules can be tracked.

And once the drift becomes visible, the conversation changes. Not because anyone suddenly becomes kinder, but because you’ve removed a form of cover.

If you can’t hide, you must explain.

Neglect: the real meaning of “98% retention”

The EVLN model’s addition of neglect is the missing piece for understanding why PBMs appear so stable.

Most employers do not exit every few years. Most do not mount sustained voice. Most renew.

In PBM slide decks, that gets framed as “98% retention,” as if it proves satisfaction and partnership.

In practice, much of that “retention” is better described as inertia.

Inertia is not endorsement. It’s exhaustion with a calendar invite.

It’s the benefits team that has been burned by an audit that went nowhere, or a renegotiation that produced a prettier guarantee but no clearer reality.

It’s the employer that’s had three account managers in two years and can’t justify the internal disruption of another switch.

It’s the pharmacy director who knows something is off but can’t get clean data fast enough to do anything about it before renewal.

It’s the consultant who senses the friction but is structurally incentivized to deliver the “safe” choice because the cost of being wrong is career-ending.

Neglect shows up as autopilot renewals, superficial quarterly reviews, and the quiet normalization of opacity. It is the passive-destructive choice to stop expecting improvement.

PBMs call this “retention.”

A more honest term is compliance-by-fatigue.

And neglect is the stable resting state of quasi-lock-in: when voice feels futile and exit feels unhelpful, staying becomes the default—not as loyalty, but as resignation.

The question that ends Part 1

If employers are stuck in quasi-lock-in—if exit is possible but functionally unhelpful—then the obvious next question is: what’s left?

Hirschman would point to voice. But in the PBM market, voice is not simply “complain more.” Voice requires a channel that can carry a signal, an audience that can act on it, and a credible consequence if the signal is ignored.

Many employers have tried voice: contract demands, audit requests, transparency riders, performance guarantees, escalation calls, “100% pass-through”.

The frustration isn’t that employers don’t speak.

It’s that the system is designed so speech doesn’t land.

Part 2 is about why voice fails in PBM relationships even when employers have leverage, consultants, and contract language on their side.

And Part 3 goes one layer deeper: if neither exit nor voice disciplines a lazy monopoly, how do you manufacture alertness? What would it mean to make exit legible—not just a switch, but a move that leaves a receipt, sharpens the signal, and forces the market to learn?

The Nigeria railway found a comfortable equilibrium: serve the captives, ignore the critics, let the demanding customers leave.

That’s not a strategy for excellence.

It’s a strategy for persistence.

The question is whether the PBM market has to remain the same way—or whether we can build enough legibility that the system can no longer survive on quiet.