AI, Ghost GDP, and Private Credit: What Markets Haven’t Priced In Yet

Markets lacking movement often hide the cause.

That’s why a recent scenario piece from Citrini Research landed with so many traders. It is written as a fictional macro memo from June 2028, looking back on how an AI productivity boom turned into a structural downturn.

The authors are clear: it is scenario work, not a forecast.

If you read it like a prophecy, it is useless. If you read it like a stress test, it is a solid prompt for real risk work.

This post focuses on two parts of that scenario that matter more than the headline numbers:

• When “friction goes to zero”: agents remove the fees and habits that many business models rely on.
• When “permanent capital” meets policyholder money: private credit problems stop being contained and start becoming a funding and confidence problem.

Both mechanisms have one uncomfortable trait: they can move through the economy before the standard data confirms anything.

When friction goes to zero, “moats” built on habit disappear

A lot of modern business is not built on better products. It is built on friction.

Friction is the small effort and time cost that stops people from price-checking everything. It includes inertia (not wanting to switch), fatigue, and “good enough” decision-making. The scenario argues that the U.S. built a large layer of profits on these human limits.

Then agents arrive, and those limits matter less.

In the scenario, agents do not wait to be asked. They run in the background based on the user’s preferences. Commerce stops being a series of human decisions and becomes constant optimization.

Once the buyer is a machine, many traditional “moats” (a company’s protection from competitors) weaken fast. A machine does not have app loyalty. It just checks options and chooses the best deal.

Interchange is the cleanest example of a friction tax

The scenario uses card payments as a simple example: if agents control more spending, they will try to route around card interchange (the fee merchants pay on card transactions).

The important point is the size of that fee. It is not small. A 2025 Federal Reserve Bank of Philadelphia working paper notes that, in the U.S., interchange fees typically range between 2% and 3% of the transaction value.

A human might accept that fee because convenience feels worth it. An agent treats that fee like a cost to remove.

Real estate shows how “relationship pricing” changes when rules and tools shift.

The scenario also argues that some industries that feel protected by “human relationships” are more fragile than people assume. Real estate is a sharp example: buyers tolerated 5–6% commissions for decades, helped by information gaps and negotiation fatigue.

You do not need to believe agents will replace realtors next year to see the direction of travel. Real estate compensation rules have already moved toward more transparency and explicit agreements.

The National Association of Realtors (NAR) said practice changes tied to its proposed settlement were implemented across the country on August 17, 2024.

NAR also explains that, as of that date, an agent working with a buyer must have a written agreement before touring a home, including key terms about compensation.

That is not an “AI claim.” It is a friction claim. AI tools would mainly speed up a process that is already becoming more negotiable and less automatic.

Why the demand hit can be bigger than the job-loss headline suggests

The scenario coins “Ghost GDP”: output that shows up in national accounts but does not circulate through households the way we are used to.

The core idea is simple. In the U.S., household spending is the main engine. When that engine weakens, a lot of earnings stories weaken with it.

The scenario’s extra point is about who gets hit first. It describes job losses concentrated in higher-income white-collar work, and it argues that this creates an outsized hit to discretionary spending because higher-income households drive a large share of total consumption.

In the real world, measuring this concentration is tricky, but the direction is clear. A Dallas Fed analysis estimates the top 20% of earners were responsible for about 57% of overall consumption on average from 2020 through Q2 2025.

The trading takeaway is not “one number is perfect.” The takeaway is that U.S. demand is concentrated enough that stress in high-income employment can matter quickly for discretionary demand, even if the total unemployment rate is not yet extreme.

“Permanent capital” is not a synonym for “no contagion”

Now the second loop, which is the part traders tend to underweight until they are forced to care.

In the scenario, private credit defaults start in software and spread into other areas. At first, the market thinks the damage should be manageable. The usual argument is: private credit is not 2008 banking. Closed-end funds have locked-up capital. Investors cannot pull money overnight.

Then the scenario turns: the system is more connected than the “locked-up capital” story suggests, partly because large alternative asset managers have integrated with insurance platforms. That matters because insurance liabilities are household money, and they sit inside a tighter regulatory and political box

The insurance integration trend is real

A 2025 research paper by Cheema-Fox, Czasonis, Kontu, and Serafeim (“Permanent Capital Meets Private Markets”) documents how life insurance platform integration has reshaped the business model and funding structure of leading alternative asset managers.

This does not mean “crisis.” It means the holder of risk changes, and the constraints change with it.

Life insurers already have meaningful private credit exposure

A 2025 Federal Reserve Bank of Chicago working paper finds life insurers increased lending in the private placement market over the past decade, reaching $849 billion (about 14% of life insurers’ balance sheets) in 2024, with private equity–owned life insurers driving a substantial part of the growth.

That is one of the non-obvious risks in the scenario: private credit is not only “private market risk.” In some cases, it is “insurance balance sheet risk.”

Openness matters because opacity delays pricing

A 2025 BIS paper on the life insurance industry flags a shift toward riskier and more opaque assets, along with related liquidity and valuation challenges that can amplify stress.

Opacity does not always cause losses. But it often delays recognition. And delayed recognition tends to create sharp repricing when the market finally agrees on the new level.

Even the Fed now frames private credit and AI sentiment as salient risks

The Federal Reserve’s November 2025 Financial Stability Report includes a survey of “salient risks” gathered from market contacts. That survey lists private credit concerns among commonly cited risks and also notes the possibility of a sharp decline in asset prices connected to a turn in AI sentiment.

That does not validate any specific scenario. It does validate that these risk categories are now treated as plausible instability triggers by real market participants.

How to think about this on a trading desk

You trade a mechanism, a timing hook, and a footprint in price.

Mechanism: what changes in cashflows or funding?

• The friction loop is a pricing-power story. Agents make it harder to charge tolls and easier for customers to switch.
• The private credit loop is a funding and confidence story. It is about marks, capital rules, and where losses sit across connected balance sheets.

Hook: what forces a repricing window?

In the scenario, the hooks are things like procurement cycles, earnings inflections, and regulation

In real markets, hooks tend to be slower and less dramatic: guidance changes, funding costs moving, credit spreads widening, and risk getting repriced in pockets before it becomes a headline.

Footprint: what does price confirm?

The footprint is rarely “one big crash” first. It is often:

• Former leaders failing to recover after good news
• Dispersion rising inside a single theme
• Credit-sensitive names rolling over while broad indices still look fine

That is not a prediction. It is a pattern traders see any time structural risk starts to matter.

Closing perspective

The scenario ends with the right reminder: you are not reading it in June 2028. You are reading it now, while the negative loops have not fully started.

The two mechanisms above are not science fiction. Pieces of them already exist:

• Large fee pools exist because humans do not optimize every purchase (interchange is one example).
• Private credit and insurance balance sheets are becoming more connected, and major institutions are openly discussing the risks that follow.

Keep your timing bar high. Size risk like you expect to be wrong sometimes. That is how an institutional process survives when narratives feel inevitable, right up until they don’t.