Citrini’s 2028 Scenario: A (Scary) Compelling Story with a Nit on Timeline

TLDR: Citrini’s 2028 scenario constructs a plausible left-tail risk — AI-driven white-collar displacement triggering a consumption collapse, private credit defaults and mortgage market stress — but compresses a structural transition that the current productivity data suggest will unfold over years not quarters. The displacement risk is also occupationally heterogeneous: workers whose expert tasks get automated face genuine downward mobility while those losing only inexpert tasks retain a skills premium. For investors, the near-term implication is not to position for imminent collapse but to be aware of holdings with moats built on white-collar income assumptions that may erode gradually but surely. The political implication is less gradual: college-educated workers experiencing unexpected downward mobility represent a new and electorally potent constituency for redistributive policy (i.e., higher taxes for the rich).

Citrini’s Projection on Employment

This past week, I came across this report from Citrini Research multiple times on my LinkedIn newsfeed. One of its central projections is that white-collar workers — roughly 50% of employment and 65-75% of discretionary spending — face mass displacement that cascades into a consumption shock outsized relative to the number of jobs lost. The mechanism is intuitive: a 2% decline in white-collar employment translates to a 3-4% hit to discretionary spending because the top income quintile drives a wildly disproportionate share of consumer activity. Displaced workers downshift into gig and service roles, compress wages in those segments too and the damage spreads economy-wide. The consumption signal lags because high earners draw down savings before behavioral change shows up in the data.

While plausible (and scary), the timeline seems to be off. The Economist’s productivity data from this week’s edition offer a useful reality check. AI added an estimated 0.25-0.5 percentage points to productivity in 2025 — essentially rounding error at the macro level. The SF Fed finds that underlying productivity gains excluding investment-driven output are close to zero. Only 13% of workers use AI daily, and AI accounts for just 5.7% of total work hours. Nine out of ten senior executives report no measurable improvement in labor productivity. The organizational rewiring — the restructuring of workflows and job definitions that would enable mass displacement — has barely begun. Historical analogues are instructive: the productivity payoff from electrification arrived decades after adoption, and the PC boom lagged similarly until firms rebuilt business models around the technology.

Another nuance the economist side of me would note: displacement risk is not uniform across white-collar work. As I noted in a previous newsletter on Autor-Thompson framework, what matters is whether AI removes the expert or inexpert tasks within a given job. A lawyer whose AI exposure is in research and synthesis faces genuine wage compression; a lawyer whose AI exposure is in document formatting and scheduling does not. The consumption shock Citrini describes would require widespread displacement of the former type, but current evidence suggests the latter is more common.

What Happens to Displaced Workers: Two Views

Citrini’s own answer is grim. Their Salesforce PM example — $180k income collapsing to $45k driving Uber — is meant to illustrate second-order math not individual tragedy. Multiply the dynamic across hundreds of thousands of workers in major metros and overqualified labor floods the service and gig economy compressing wages for workers already at the margin. The authors go further by closing the escape route: autonomous vehicles eventually displace the gig economy that absorbed the first wave. In this view, reabsorption fails because AI capability improves faster than retraining can respond.

A more nuanced read applies the same logic directly to Citrini’s own example. The Salesforce PM earning $180k is not a monolithic role — it bundles expert tasks like product strategy, stakeholder alignment and ambiguous tradeoff resolution alongside inexpert ones like drafting status updates, formatting roadmaps and compiling meeting notes. If AI automates the latter the PM likely keeps their job, with work concentrating around the higher-skill residual. The grim outcome — the $45k Uber shift — only materializes if AI reaches the expert task bundle too. That is the scenario worth stress-testing, but it is further out on the capability curve than Citrini’s timeline implies and far from uniform across PM roles. The more probable near-term outcome for many white-collar workers is not displacement but a quieter erosion: fewer headcount additions, and a widening gap between those whose remaining task bundle commands a premium and those whose does not.

The political implication deserves attention, especially for high-income workers. A large constituency of college-educated workers experiencing unexpected downward mobility — a group historically insulated from automation — would generate meaningful electoral pressure for redistributive policy. A compute tax or sovereign claim on AI-generated output (proposals already surfacing in Citrini’s scenario) becomes more politically viable well before 2028 if white-collar job losses accelerate even modestly.

In short, I wouldn’t panic but prepare (by thinking carefully about the type of tasks you carry out and the degree of human judgment required in each).