What is the lemons problem in economics?

The lemons problem, described by George Akerlof in 1970, is what happens when one side of a market has better information than the other about the quality of what is being sold. Buyers, unable to tell good from bad, offer only an average price, which drives high-quality sellers out of the market. As quality exits, the average drops, prices fall further, and the cycle repeats until only the worst participants remain. Akerlof received the 2001 Nobel Prize in Economics partly for this insight.

Why is adverse selection especially bad in medical tourism?

Medical tourism layers multiple information gaps on top of each other. The patient knows less than the physician, who knows less than the hospital, who knows more than the platform, which knows more than the agent. The outcome of a procedure emerges weeks or months later, long after the transaction has closed and the patient has returned home. And low-quality hospitals spend heavily on marketing while high-quality ones do not, so visibility itself becomes biased toward worse providers.

What is the difference between a marketplace and a trust protocol?

A marketplace is an open venue where any qualifying supplier can list and compete, typically on price and visibility. A trust protocol is a structured set of rules that control who can participate, under what verification standards, and with what long-term accountability. The difference matters because a marketplace must assume participants will sort themselves fairly, while a trust protocol does not rely on that assumption — it uses mechanism design to make high-quality participation economically rational and low-quality participation economically difficult.

How does AetherHeal prevent adverse selection?

Four design decisions work together. First, participation requires a costly verification process that low-quality hospitals cannot sustain. Second, visibility inside the platform is decoupled from marketing spend, so no hospital can buy its way to the top. Third, patient reviews are replaced by expert physician-led screening against clinical standards. Fourth, outcomes are tracked long after discharge, which turns a one-shot transaction into a repeated game where reputation becomes a durable asset.

What is costly signaling, and why does it matter for platforms?

Costly signaling, introduced by Michael Spence in 1973, is the mechanism by which a high-quality participant voluntarily takes on a cost that a low-quality participant cannot afford to imitate. On a medical platform, a deliberately difficult verification process — clinical documentation, credential review, outcome history, direct physician interviews — acts as a signal precisely because low-quality hospitals cannot pass it without being filtered out. If verification becomes cheap or automatic, it stops being a signal.

Why are reviews unreliable as a quality signal in medical tourism?

Medical reviews are collected in a narrow post-procedure window when the patient is still emotionally committed to the decision they just made, before long-term outcomes become visible. They can be gamed by clinics that solicit positive reviews selectively. They confuse customer satisfaction with clinical quality, which are different things. And they rely on the patient's ability to judge clinical work, which is exactly the information asymmetry that causes the lemons problem in the first place.

What are the remaining risks AetherHeal has not solved?

Three. Verification capture: if the verification layer becomes economically dependent on the hospitals it evaluates, it becomes a rubber stamp. Survivorship illusion: if every verified hospital receives the same label, the label loses its signal value, so re-verification and performance-based tiering are required. Cold start: building the first cohort of high-quality hospitals is genuinely hard because the chicken-and-egg problem overlaps with the lemons problem — early recruitment must happen through direct relationships rather than commission-based economics.

What should patients ask before choosing a medical tourism platform?

Two questions. First: what did the hospitals have to do to get on this platform, and is the answer more than a registration form? Second: what happens if a hospital's outcomes deteriorate after it joins — does the platform have a mechanism to remove it, and what is the feedback loop that triggers that mechanism? A platform that cannot answer both questions with specifics is running an open marketplace vulnerable to adverse selection, regardless of what its marketing claims.

Back to Blog

Insights

Why AetherHeal Is Not a Marketplace

A 1970 Nobel Prize paper on used cars explains why open medical marketplaces decay — and why AetherHeal was designed as a trust protocol instead.

Dr. Jee Hoon JuApril 12, 202611 min read

Key Takeaways

Open medical marketplaces are vulnerable to the same decay mechanism George Akerlof described in his 1970 Nobel Prize paper on used cars: information asymmetry causes bad suppliers to drive out good ones.
In medical tourism the information gap is larger and feedback is delayed by months, which makes the decay faster and harder to detect than in consumer markets.
The economic mechanisms that block adverse selection are costly signaling, screening, bonding, and repeated games — not reviews, ratings, or price competition.
AetherHeal is structured as a trust protocol rather than a marketplace because verification cost, decoupled visibility, physician-led review, and long-term outcome tracking are the only durable ways to prevent the lemons problem.
The practical test of any medical platform is not how many hospitals are listed but what they had to do to get on the list and what happens if they stop deserving to be there.

The Nobel Prize That Explains Why Medical Marketplaces Fail

In 1970, an economist named George Akerlof published a short paper about used cars. It earned him a Nobel Prize.

The paper was not really about cars. It was about what happens when one side of a transaction knows more than the other — and the buyer can't tell the difference before paying. Akerlof showed, with devastating simplicity, that when information is asymmetric in this specific way, the bad drives out the good. Markets that look open and competitive quietly decay until only the worst participants remain.

This insight is the single most important thing I thought about when deciding what AetherHeal would and would not be.

It is also the reason AetherHeal is not a marketplace.

Akerlof's Lemons Problem, in Plain English

Imagine a used car market where sellers know whether their car is a peach or a lemon. Buyers don't. The buyer, unable to tell one from the other, rationally offers the average price — the price of a medium-quality car.

That average price is great if you have a lemon. You get more than it's worth. But if you own a peach, the average price is an insult. So you keep it off the market.

Now the average quality of cars for sale has dropped — the peaches are gone. Buyers notice, and revise their offer downward. The next tier of decent cars now finds the price insulting, and exits. The cycle repeats. Each iteration raises the concentration of lemons and lowers the price until, in Akerlof's model, the market can collapse entirely.

The formal name is adverse selection. The informal name is the lemons problem. The mechanism is always the same: information asymmetry → rational exit by quality → declining average → further rational exit → collapse.

It is one of the few results in economics that makes you look at a perfectly functional-looking market and realize it is quietly eating itself.

How This Plays Out in Medical Platforms

Now replace "used cars" with "hospitals" and "buyers" with "international patients." The mechanism is not just similar — it's stronger, because the information asymmetry is worse.

Consider what happens when an open medical tourism marketplace launches and asks hospitals to list.

Which hospitals join first? Not the ones with established reputations and full waiting rooms. Those hospitals don't need a platform — they have their own channels, existing referrers, and a brand that acquires patients on its own. Paying a marketplace commission is, for them, an unnecessary cost.

The hospitals that sign up first are the ones that need patients. That is not the same as the ones that deserve them. The platform's initial supply pool is structurally biased toward the bottom of the quality distribution. Not because the founder intended it. Because rational economic agents sort themselves that way.

Patients then arrive, experience the average of what the platform offers, and tell their friends. The discerning patients — the ones who ask good questions and choose carefully — leave fastest, because they are the ones who most quickly recognize the pool they are fishing in. What's left is a pool of patients who either can't tell the difference or have no alternatives. That's a pool no quality hospital wants either.

The spiral is now complete: worse hospitals → worse patient experiences → better patients leave → better hospitals leave → worse hospitals still.

I've watched this happen in real time to every "medical tourism marketplace" I've evaluated as a physician. The surface explanations vary — "the reviews got gamed," "the big hospitals wouldn't join," "we had a bad cohort of patients this quarter." The structural explanation is always the same one Akerlof wrote down in 1970.

Why This Is Worse in Medical Tourism Than Anywhere Else

Three things make adverse selection especially vicious in medical tourism specifically:

The asymmetry is asymmetric. In a normal marketplace, the seller knows more than the buyer. In medical tourism, the hospital knows more than the physician, who knows more than the platform, who knows more than the agent, who knows vastly more than the patient. It's not a single information gap — it's a layered chain, where each step compounds the disadvantage. The patient at the end of that chain has no language, no medical context, no local legal recourse, and no way to verify what they're being told.

Results arrive late. You can test-drive a used car. You cannot test-drive a facelift, a hair transplant, or a spine fusion. The real outcome emerges weeks or months later — after the patient has flown home, the platform has been paid, and the reviews have already been written in the narrow window of post-procedure optimism. Structural follow-up is rare. Retrospective accountability is rarer still.

Marketing budget becomes a substitute for quality. Hospitals with clinical confidence don't advertise aggressively; they don't need to. Hospitals that need patients spend heavily on digital marketing, paid placements, and commission-based referrals. On any open platform where visibility correlates with spend, the same mechanism Akerlof described plays out in the visibility layer itself: the hospitals patients see are not the hospitals they should see.

This is adverse selection with the volume turned up. Same theorem, worse parameters.

What Economics Offers as Solutions

After Akerlof's paper, economists spent the next two decades figuring out how to prevent the lemons problem when it threatens a market worth preserving. Four mechanisms emerged:

Signaling (Spence, 1973). A high-quality supplier voluntarily pays a cost that a low-quality supplier cannot afford to imitate. A diploma from a hard university. A long warranty on a product. A willingness to stake something real on being right. The signal works precisely because it's expensive — and more expensive for the impostor than for the genuine article.

Screening (Stiglitz, 1981). The informed side of the market designs options that force the other side to self-sort. Insurance companies offer bundles where healthy and sick customers rationally pick different plans. The bundles don't need to verify who is who — the choice itself reveals it.

Bonding and warranties. Sellers post a bond that they lose if quality fails. Only sellers confident in their quality are willing to post the bond. The willingness is the quality signal.

Reputation and repeated games. When a supplier's long-term payoff depends on past behavior, short-term cheating becomes irrational. Reputation becomes an asset worth protecting. This one requires time: it doesn't work on day one.

Every serious attempt to solve the lemons problem uses some combination of these. They are not clever UX tricks. They are mechanism design — shaping the structure of the transaction so that quality is revealed through the shape itself, not through hopeful trust.

How AetherHeal Is Designed to Block Adverse Selection Structurally

This is the part of the AetherHeal design that looks, from the outside, like ordinary product decisions — and from the inside, like every single one of them is answering Akerlof.

Participation has a real cost, and it is not money. Hospitals that want to be part of AetherHeal go through a verification process that is deliberately non-trivial: clinical documentation, credential review, outcome history, direct physician interviews. It takes weeks. It has to, because the entire point of the cost is that a low-quality hospital cannot sustain it without being filtered out. This is costly signaling in its purest form.

Visibility is decoupled from marketing spend. On AetherHeal, a hospital cannot buy its way to the top of a patient's recommendation. The ranking is produced by the verification layer, not by the ad auction. If we allow the marketing-budget-to-visibility channel to re-open, Akerlof's mechanism resumes immediately. Holding that line is not a policy choice; it is a structural requirement.

Reviews are replaced by expert screening. Patient reviews on medical procedures are systematically unreliable — collected too early, too emotional, and too easy to game. AetherHeal uses a physician-led review layer (what we internally call the Angel Physician) to examine cases against clinical standards, not against customer satisfaction. This is Stiglitz-style screening: the informed party designs the evaluation, not the uninformed one.

Long-term outcomes are tracked, not abandoned. The hardest part of medical tourism is that the feedback loop closes after the patient goes home. AetherHeal's architecture includes structured post-procedure follow-up so that a hospital's reputation is connected to what actually happened, not to the moment of discharge. This is what turns a one-shot transaction into a repeated game — and once you have a repeated game, reputation becomes an asset worth defending.

Pull any of these four pieces out, and adverse selection starts working on the platform the moment it goes live. Keep all four in place, and something different becomes possible: a structure where the rational move for a good hospital is to participate, and the rational move for a bad hospital is to stay out.

That structure is not a marketplace. A marketplace is an open stage where anyone can perform. What AetherHeal builds is closer to a trust protocol — a set of rules that shape who can participate, on what terms, and with what accountability. The distinction is not semantic. It is the difference between a market that decays and one that doesn't.

What Still Keeps Me Up at Night

I don't want to overstate any of this. Mechanism design is not a solved problem, and AetherHeal is not immune to adverse selection forever. The three risks I think about most:

Verification capture. If the verification layer ever becomes economically dependent on the hospitals it's verifying, it stops being a filter and becomes a rubber stamp. This is the same failure mode that hits credit rating agencies, drug safety boards, and any institution asked to grade the people who pay it. The defense is strict separation: the incentives of the verifiers must never be tied to the outcomes of the verified.

Survivorship illusion. Even inside a verified pool, there is a quality distribution. If every verified hospital gets the same label, that label loses its signal value. Re-verification and performance-based tiering are necessary to keep the signal from going flat.

The cold start. The mechanism design solutions to adverse selection all presume the platform exists. Getting from zero to a critical mass of high-quality supply is genuinely hard, because the chicken-and-egg problem overlaps with the lemons problem. The only way through this early phase, as far as I can tell, is to bootstrap on direct relationships rather than transactions — to recruit the first hospitals on vision and alignment, not on commission economics, and to use their presence as the social proof that makes the next cohort easier.

Why This Matters Before You Choose Any Platform

If you are a patient reading this, the reason it matters is direct. The platform you choose is not a neutral pipe through which information flows. It is a filter. What reaches you has already been shaped by that filter — by whose incentives it serves, what it excludes, what kinds of quality it can and cannot signal. When you evaluate a platform, the question is not "how many hospitals do they list?" The question is "what did they have to do to get on that list, and what happens if they stop deserving to be on it?"

A marketplace that cannot answer those two questions well is a lemons market in slow motion. A protocol that can answer them is something else entirely — something where the structure itself is the quality guarantee, not a promise on top of it.

AetherHeal is built around that second thing. Not because it's a nicer story, but because the first thing has a fifty-year-old Nobel Prize explaining why it can't work.

For the foundational theory behind this reasoning, Akerlof's original 1970 paper "The Market for Lemons" is freely available through the PubMed and economics literature on adverse selection in healthcare markets — a useful entry point if you want to see how the same mechanism has been applied to insurance, pharmacy benefits, and physician networks over the decades since.

Dr. Jee Hoon Ju is a physician and the founder of AetherHeal, a physician-led trust protocol for international patients considering medical care in Korea. Read more about how AetherHeal is designed to protect patient trust, or explore why AI cannot replace physicians in healthcare.

Frequently Asked Questions

What is the lemons problem in economics?: The lemons problem, described by George Akerlof in 1970, is what happens when one side of a market has better information than the other about the quality of what is being sold. Buyers, unable to tell good from bad, offer only an average price, which drives high-quality sellers out of the market. As quality exits, the average drops, prices fall further, and the cycle repeats until only the worst participants remain. Akerlof received the 2001 Nobel Prize in Economics partly for this insight.
Why is adverse selection especially bad in medical tourism?: Medical tourism layers multiple information gaps on top of each other. The patient knows less than the physician, who knows less than the hospital, who knows more than the platform, which knows more than the agent. The outcome of a procedure emerges weeks or months later, long after the transaction has closed and the patient has returned home. And low-quality hospitals spend heavily on marketing while high-quality ones do not, so visibility itself becomes biased toward worse providers.
What is the difference between a marketplace and a trust protocol?: A marketplace is an open venue where any qualifying supplier can list and compete, typically on price and visibility. A trust protocol is a structured set of rules that control who can participate, under what verification standards, and with what long-term accountability. The difference matters because a marketplace must assume participants will sort themselves fairly, while a trust protocol does not rely on that assumption — it uses mechanism design to make high-quality participation economically rational and low-quality participation economically difficult.
How does AetherHeal prevent adverse selection?: Four design decisions work together. First, participation requires a costly verification process that low-quality hospitals cannot sustain. Second, visibility inside the platform is decoupled from marketing spend, so no hospital can buy its way to the top. Third, patient reviews are replaced by expert physician-led screening against clinical standards. Fourth, outcomes are tracked long after discharge, which turns a one-shot transaction into a repeated game where reputation becomes a durable asset.
What is costly signaling, and why does it matter for platforms?: Costly signaling, introduced by Michael Spence in 1973, is the mechanism by which a high-quality participant voluntarily takes on a cost that a low-quality participant cannot afford to imitate. On a medical platform, a deliberately difficult verification process — clinical documentation, credential review, outcome history, direct physician interviews — acts as a signal precisely because low-quality hospitals cannot pass it without being filtered out. If verification becomes cheap or automatic, it stops being a signal.
Why are reviews unreliable as a quality signal in medical tourism?: Medical reviews are collected in a narrow post-procedure window when the patient is still emotionally committed to the decision they just made, before long-term outcomes become visible. They can be gamed by clinics that solicit positive reviews selectively. They confuse customer satisfaction with clinical quality, which are different things. And they rely on the patient's ability to judge clinical work, which is exactly the information asymmetry that causes the lemons problem in the first place.
What are the remaining risks AetherHeal has not solved?: Three. Verification capture: if the verification layer becomes economically dependent on the hospitals it evaluates, it becomes a rubber stamp. Survivorship illusion: if every verified hospital receives the same label, the label loses its signal value, so re-verification and performance-based tiering are required. Cold start: building the first cohort of high-quality hospitals is genuinely hard because the chicken-and-egg problem overlaps with the lemons problem — early recruitment must happen through direct relationships rather than commission-based economics.
What should patients ask before choosing a medical tourism platform?: Two questions. First: what did the hospitals have to do to get on this platform, and is the answer more than a registration form? Second: what happens if a hospital's outcomes deteriorate after it joins — does the platform have a mechanism to remove it, and what is the feedback loop that triggers that mechanism? A platform that cannot answer both questions with specifics is running an open marketplace vulnerable to adverse selection, regardless of what its marketing claims.