The Unanchored Central Banker: Manoj Pradhan on Inflation, Demographics, and Why AI Won't Save Us

ELI5/TLDR

Manoj Pradhan returns for a follow-up to his 2020 book with Charles Goodhart, this time arguing that the demographic squeeze isn’t only inflationary — it’s also going to break central banks. As populations age, governments will borrow more to pay for pensions and care, China’s deflationary export pipeline runs out, and central bankers will lose the political cover they had in the easy decades. The “unanchored central banker” is one who can no longer fight inflation single-mindedly because the government won’t let them.

The Full Story

The sequel question: did demography still call inflation?

In 2019, Pradhan and Goodhart wrote The Great Demographic Reversal arguing that the secular disinflation of the 1990–2020 era was about to flip. The book landed in Q1 2020, just as the world was preparing for another deflation. They were right. The sequel, The Unanchored Central Banker, picks up the same thread and asks: now that the inflation thesis has played out, what comes next for real rates, fiscal policy, and the central banks that have to clean it all up?

The short answer is that the structural story has gotten worse, not better. AI doesn’t fix it. China doesn’t fix it. And the central bank’s traditional toolkit becomes politically unworkable somewhere along the way.

Why aging societies don’t save you out of the problem

The standard model says: people live longer, so they save more, so real rates fall. Pradhan accepts the logic but argues the model is missing a leg. The real rate is a price that clears private surplus against public deficit. As societies age, the public deficit blows out — pensions, healthcare, dementia care, all of it scales with the elderly share. The CBO and OBR forecasts, in his view, systematically understate this.

“The public sector deficit is going to rise by much more than the private sector savings, which means that the real interest rate is going to rise, not fall over the decades to come.”

That’s the whole rate story in one sentence. Saving more doesn’t help if the state is borrowing even harder.

The dual Phillips curve

This is Pradhan’s pet idea and the engine of the book. Inflation isn’t one thing — it’s two:

Goods inflation has been globally determined since roughly 1990, and overwhelmingly by China after WTO accession in 2000. Washing-machine prices ran from +3% in 1990 to minus 2% before the GFC. That negative goods inflation acted like a quiet income transfer to Western consumers.
Services inflation is domestic. Wages, rents, haircuts, healthcare. Central banks can actually move this, but only by hurting the labour market.

For thirty years, central banks looked competent because China was doing half their job. The Bank of England didn’t crush services inflation; falling TV prices made the headline number look fine. Strip out the goods tailwind and the picture is much less flattering. Going forward, China’s deflation impulse fades as its own demography turns and its current account surplus eventually flips. Which leaves central banks with services inflation, the labour market, and no Chinese subsidy.

The China shock 2 — strategy or accident?

The fashionable read is that Beijing has chosen export deflation as an employment policy: can’t stimulate consumption, so dump cheap EVs and solar panels on Europe instead. Pradhan thinks this gets the causality backwards.

The deflation, in his view, is mostly domestic. The property bust they tried and failed to engineer in 2012–15 and then re-inflated has finally arrived. Subsidies in EVs, batteries and solar created brutal internal price wars that bled out into export prices. Policymakers don’t like this — China has a debt mountain, and persistent deflation makes that mountain heavier. And manufacturing employment in China has been falling for decades; the marginal employment is in services. So the idea that Beijing is consciously running factories flat-out to soak up labour doesn’t quite stack up. The exports are a symptom, not the strategy. Which also means the impulse fades on its own timeline, not on a tariff timeline.

Why AI doesn’t save you

Pradhan does the rare thing for a macro forecaster: he refuses to make a confident AI prediction. But he points out three things people miss.

First, the capex bill. Over $600bn in US AI capex is competing for capital at the same moment that US debt-to-GDP has crossed 100%. Even if the spending is profitable, it puts upward pressure on real rates.

Second, electricity. Data-centre power demand layered on top of normal demand pushes energy prices into everything.

Third, and most importantly: AI is currently a cognitive-work technology. The labour shortage in an aging society is mostly physical — someone has to lift the elderly relative, change the dressing, drive the van. Those workers get pulled out of productive capacity into care, and AI doesn’t substitute for them. Which means AI accelerates the productivity of the workers who matter least for solving demography, while the demographic constraint binds anyway.

“Those workers are then taken away from what you and I would call productive work for the economy… They’re doing a very, very, very rewarding social task, but that social task doesn’t contribute to future productivity growth.”

If AI does cause a big collapse in cognitive-labour demand, Pradhan thinks UBI becomes politically inevitable, financed by taxing the AI-winner companies — which then claws back a chunk of the productivity gain. Either way, no free lunch.

On inequality he is mildly more optimistic. AI compresses the gap because a median worker getting a 30% productivity bump matters far more in aggregate than an elite worker getting 50%, simply because there are twenty times as many of them. Same logic for small firms acquiring CFO-equivalent capabilities, and for emerging markets leapfrogging the “education, education, education” bottleneck.

The unanchored central banker

This is the title and the punchline. The book’s argument is not that politicians are bullying central banks. It’s that the conditions which made central banks look heroic are gone.

For three decades, every cut was a gift. Borrowing got cheaper, house prices rose, governments got re-elected, central bankers got knighthoods. Now the direction reverses. Aging-driven deficits keep rising. About 70% of US debt is financed at the short end, so every Fed hike feeds directly back into the deficit, which then feeds back into inflation expectations.

“You can’t really fight inflation aggressively in a highly indebted society. And that puts the central bank in conflict with the government.”

The political solution is to appoint central bankers willing to live with the government’s fiscal stance. Which, combined with a shrinking Fed balance sheet (Warsh’s stated preference) and ballooning issuance, hands ever more duration to private investors. Pradhan thinks that gap eventually forces the balance sheet to start growing again — not because anyone wants it to, but because the alternative is a disorderly market.

Bessent’s strategy of skewing issuance to the short end is, in this frame, a directional bet that long rates fall. Pradhan and Boyle both think that’s the wrong bet. The credible way to bring borrowing costs down, learned the hard way in every emerging market, is to shrink the primary deficit — actual new spending versus revenue. Tinker with the interest line and markets see through it.

“If your primary deficit goes down… markets give you the benefit of the doubt.”

Truss’s gilt episode, the recent French fiscal scare, the Japanese election wobble — these are the early-warning lights. Advanced economies have been immune to bond-market discipline for so long that everyone forgot what it looks like.

Key Takeaways

The real-rate model that says “aging → more saving → lower rates” misses the public-deficit side, which scales harder with demographics than private savings do.
Goods inflation has been globally determined since ~1990; China’s WTO accession in 2000 turned it negative. Strip that out and central bank performance from 1990–2020 looks much less impressive.
Goods inflation around 1990: +3%. Just before the GFC: -2%. That’s the size of the tailwind central banks had.
China’s current “deflation export” is mostly a spillover from the domestic property bust plus subsidised-sector price wars in EVs/batteries/solar — not a deliberate employment strategy. Manufacturing employment in China has been falling for decades anyway.
AI capex in the US is north of $600bn at a time when debt/GDP just crossed 100%. Even if profitable, this pushes real rates up.
The demographic labour shortage is physical-care work, not cognitive work. AI augments the wrong workers for the problem we have.
If AI does collapse cognitive labour demand, UBI becomes politically inevitable and gets financed by taxing AI winners, eating into the gain.
AI is probably inequality-reducing in aggregate because the median worker matters more macro-economically than the elite worker (40% of the labour force × 30% productivity gain >> 2% × 50%).
US Social Security trust fund runs out around 2033. Politicians will almost certainly bankroll the gap via debt rather than cut benefits — Pradhan’s “cynical guess”.
About 70% of US government debt is financed at short rates, so every Fed hike directly worsens the deficit. This is the fiscal-monetary loop that disarms aggressive hiking.
The “unanchored” central banker is unanchored not by political pressure but by the impossibility of fighting inflation while the government is running expansionary fiscal — financial markets will demand a higher term premium regardless.
Warsh’s stated preference for a smaller Fed balance sheet is, in Pradhan’s view, probably untenable: rising issuance plus shrinking central-bank demand means private investors absorb a gap that grows without bound.
The credible playbook for lowering borrowing costs (from every EM experience) is shrinking the primary deficit, not engineering the interest line.
Early bond-market warning signals to watch: the 2022 Truss episode, the French fiscal scare, and the Japanese election fiscal wobble. Advanced economies have been immune to discipline; that immunity is fading.

Claude’s Take

This is a high-density interview where two finance-literate people talk past nothing. Patrick Boyle knows the material well enough to push, and Pradhan has had six years to refine the thesis under live-fire conditions where his 2019 call paid off. The track record buys him a hearing, and most of what he says holds up.

The strongest parts are the dual Phillips curve and the fiscal-monetary feedback loop. Both are framings that, once you see them, you can’t unsee. The observation that 70% of US debt is short-end financed and therefore every hike worsens the deficit is one of those quiet structural facts that explains a lot of central-bank behaviour without needing any conspiracy. The China-deflation-as-symptom-not-strategy reading is also more careful than the standard “China shock 2” narrative.

Two soft spots worth flagging. First, the AI section is honestly hedged (“we’re not technology experts”) but in places leans on intuitions about cognitive vs physical labour that may not survive another two years of robotics progress. Pradhan basically concedes this. Second, the inequality argument — that AI helps the middle 40% more than the top 2% — is a real possibility, but the empirical evidence is still thin and the distributional politics of UBI are doing a lot of work in the model.

The interview is also notably short on confident predictions about the cyclical picture. Pradhan keeps saying “we can have cyclical episodes” and “we’ll have to see how the labour market evolves” — which is honest, but means this is more a structural-framework conversation than a trade idea.

Score 9. Specific, well-organised, contrarian where it should be and hedged where it should be. The dual Phillips curve alone is worth the hour. Half a point off only because there’s some repetition with the first book if you’ve read it.