Central Banks Are Losing Control | Ideas Lab

ELI5/TLDR

For about thirty years, central banks had an easy job. A flood of cheap labour — baby boomers, women joining the workforce, and above all China — kept the price of goods falling, which kept inflation low almost by itself. That flood is now drying up. The world is ageing, birth rates have collapsed almost everywhere, and immigration is being shut off. Manoj Pradhan argues this flips the whole picture: labour gets scarce, governments have to spend more and more on pensions and elderly care, debts balloon, and inflation stops being a gift from China and starts being a domestic problem. The punchline: central banks won’t be able to fight inflation the way Paul Volcker did, because raising rates hard now would blow up government finances. So when inflation and debt stability collide, they’ll quietly choose to protect the debt — and let inflation run a bit hotter.

The Full Story

Manoj Pradhan runs a macro consultancy called Talking Heads Macro, and he co-wrote two books with Charles Goodhart, one of the more respected names in monetary economics. The first, The Great Demographic Reversal (2020), made an unfashionable bet: that decades of falling inflation were about to reverse. Much of that has since gone mainstream. This conversation is about the sequel, The Unanchored Central Banker, which follows the demographic argument to its uncomfortable conclusion.

The sweet spot we just left

Start with why the good times were good. Pradhan describes a “demographic sweet spot” built on three forces: baby boomers entering work, women joining the workforce, and China (plus Eastern Europe) being absorbed into the global economy.

The key word is supply. All three dumped an enormous amount of labour into the world at once. More workers meant cheaper labour, which meant cheaper goods. China mattered most because its workforce wasn’t just cheap — it was skilled at every level, able to compete in basic manufacturing and cutting-edge technology alike.

“When a workforce like that joins the global labour force, the supply shock is felt across a range of products.”

Cheaper labour also dragged down interest rates. Capital flowed out of the West and into China, which lowered the amount of investment chasing savings at home — and the “real” interest rate (the rate stripped of inflation) is essentially set by that tug-of-war between savings and investment. More savings, less investment, lower rate.

The result was a near-magical environment. Falling inflation, falling rates, rising house prices, cheap government borrowing. Pradhan’s host calls it the honeymoon period of a marriage, and it fits: governments happily granted central banks independence because the arrangement made incumbents easy to re-elect. Everyone got what they wanted.

Why all three engines are stalling

Women’s participation in advanced economies has risen to nearly match men’s — there’s little headroom left. The boomers are retiring, not joining. And China’s labour supply is “dwindling to the extent that cities within China themselves are in intense competition for the next round of migrant workers.”

Then there’s the birth rate. The replacement rate is 2.1 children per woman; almost every country except parts of sub-Saharan Africa is now below it — including India at the national level, something that wasn’t true four years ago. The causes are stubborn. Healthcare improvements mean families no longer have many children as insurance against some dying young. And surveys consistently point to the cost of raising children, plus careers, as the reasons people stop at one or none. Governments have tried to bribe their way out — Hungary, China — with little to show for it.

“Something that gets them above 2.1 looks incredibly difficult to me.”

The one post-pandemic surprise — a global immigration surge — is now reversing too, as political backlash slams the door in country after country. So the obvious release valve (import workers from places that still have a surplus) is closing just when it’s needed.

Here’s where Pradhan parts ways with conventional models. Most economists accept the population is ageing but don’t see how that pushes interest rates up. He says they’re missing two things.

Blind spot one: the government. The standard counterargument is that longer lifespans make people save more for a longer retirement, and all that saving floods the bond market and pushes rates down. The logic is sound — as far as it goes. But households aren’t the only player. Governments have to respond to an ageing society by spending vastly more on pensions and healthcare, especially as expensive conditions like dementia pile up.

“There is an undeniable fact facing us that government dis-saving is probably going to rise even faster.”

Put the two together and government dis-saving overwhelms household saving. A net dis-saving economy is one where real interest rates can’t stay low. Most models dodge this by assuming government deficits and pensions stay a fixed share of GDP — an assumption Pradhan says simply breaks once you take ageing seriously.

Blind spot two: housing. Messier, but real. Population growth drives demand for homes, and homes are most families’ biggest asset. The wrinkle is that ageing reduces the fungibility of housing — old people don’t move. A young family throws everything in a U-Haul and goes; an 80-year-old, disoriented away from a familiar room, doesn’t. So old homes don’t free up for the young, and new family formation needs new homes built. Add rural depopulation (Italian towns selling mansions for a euro, people clustering into cities for healthcare) and you get persistent construction demand — more upward pressure on real rates.

Does AI rescue us?

If the problem is too few workers, isn’t AI the answer? Pradhan is careful here, and the care is the point. He uses Claude daily, built a fiscal-policy simulator with it (“it took me twenty times more to make sure the data was right than it took Claude to build the simulator”), and respects the disruption. But he flags three things the mainstream conversation misses.

First, the doomsday case. If AI really did collapse the demand for labour, you’d need something like universal basic income — and you couldn’t fund it by taxing society at large; politically impossible. You’d have to tax the companies that shed workers while keeping profits. Which means the very firms investors are betting on — the ones replacing labour — would be the ones taxed to pay for it. The trade may not work out the way the market assumes.

Second, AI’s shock is spread out. Past technology shocks hit manufacturing, which clusters in places (Sheffield steel, Detroit cars) or sectors, creating concentrated pools of unemployment that are hard to redeploy. AI hits services, which are everywhere. Cut 20% of accounting work and you’ve trimmed a slice from every town and firm rather than gutting one region — easier to absorb, and the freed-up money can flow to a company’s core work. As with China, a shock that destroys some jobs can, via cheaper everything, create more jobs elsewhere that nobody can predict in advance.

Third — his most controversial claim — AI may reduce inequality, and that pushes rates up. AI knocks down the barriers guarding high-paid service jobs (law, accounting), opening them to more people. Ignore the top fraction of a percent who get spectacularly rich, and on an 80/20 view inequality could fall. And there’s a paper (Mian, Straub, Sufi) arguing that higher inequality means lower rates, because the rich save more and that saving depresses rates. Flip it — less inequality, less of that saving glut — and rates drift higher.

And even on its own terms AI isn’t frictionless: hyperscaler capex has breached $700 billion, mostly paid in cash pulled out of money-market funds. That’s a double whammy for rates — you remove loanable funds and spend them on capital. Layer on roughly $600 billion the “big beautiful bill” adds to this year’s deficit and you get a ~$1.3 trillion swing in demand for funds in a single year. Plus the electricity and data-centre demand AI itself creates. None of it points to lower rates.

The unanchoring

Now the title. Pradhan splits the famous Phillips curve — the relationship between economic slack and inflation — into two. The services curve is the classic one: hot economy, higher wages, higher services prices, generated at home. The goods curve, since the 1990s, has had almost nothing to do with the domestic economy — goods prices are set largely in China and in commodities. The disinflation of the past three decades came from goods. Central banks barely had to do anything; they reaped the benefit.

The future is the opposite. Goods disinflation is over; what matters now is services inflation, which is tied to labour markets — and labour is getting scarce while governments spend more, keeping markets tight. So inflation becomes domestic, sticky, and structural.

This is where governments trap the central bank. Compare Paul Volcker, who broke 1970s inflation by ramming rates sky-high and forcing two recessions. He could do that because there was little debt around — the pain hit the economy, not fiscal stability. Try it today against a 100%+ debt-to-GDP ratio and double-digit rates would explode the cost of financing the debt. You might win the inflation battle this year, but the deficit you create becomes debt next year, which you’ll then need future inflation to dissolve.

“You are no longer, as a central banker in a highly indebted society, able to single-mindedly pursue inflation. You will have to take some choice between fighting inflation and balancing off fiscal instability.”

When that choice comes, Pradhan thinks supporting the bond market wins and the inflation target slips. He points to the bond-market reactions that hit the UK’s Truss budget, France, Japan, and even the US — the kind of vigilante response once reserved for emerging markets. The “advanced economy gets a free pass” era is ending. His sharpest analogy is Brazil: its primary deficit (new spending) is under half a percent of GDP and within its own fiscal rule, yet its total deficit runs 8–9% — almost all of it interest, because the central bank cut too early, let inflation up, and now can’t escape sky-high rates. Advanced economies won’t hit 14–15% policy rates, but at 100–140% debt-to-GDP, even an extra two or three points changes everything.

The escape hatches, and why he doubts them

The host pushes back: in a leveraged world, won’t yields above ~5% trigger a funding crisis, forced deleveraging, recession — which drags rates back down on its own? Pradhan grants some validity (the UK is living a version of it) but adds a darker wrinkle: governments can respond to high borrowing costs not by behaving, but with financial repression — leaning on central banks until they cave and cut. He doubts anything changes smoothly without a genuine crisis, because the two real fixes — cutting non-discretionary spending (mostly healthcare) and resisting the urge to splurge before elections — are both obvious and both nearly impossible. There are honourable exceptions (South Africa’s central bank quietly forcing austerity via a lower inflation target; New Zealand chasing a primary surplus) but they prove the rule.

His proposed fixes are creative and politically awkward: tax incentives to hire workers over 55 (graded by how cognitive the job is, since judgement improves with age even as the body fails), flexible retirement ages, and — most provocatively, framed as pure economics with no moral judgement — a slightly higher tax on people who choose not to have children, ring-fenced in escrow to fund their own future healthcare. The logic: families provide the overwhelming majority of elderly care, so the childless shift that burden onto the state and onto other people’s children.

Key Takeaways

The 30-year disinflation was mostly imported. It came from falling goods prices set in China and commodities, not from central-bank skill. That tailwind is gone.
Real interest rates are set by savings vs. investment. China pulled investment out of the West, lowering the equilibrium rate. Reverse the labour glut and the mechanism reverses.
The replacement fertility rate is 2.1. Nearly every country except parts of sub-Saharan Africa is now below it — including India nationally. Pro-natalist bribes (Hungary, China) have largely failed.
The key ageing blind spot is the government, not the household. Longer lives may raise household saving, but government dis-saving (pensions, healthcare, dementia care) rises faster. Net result: upward pressure on rates.
Housing ages badly. Old people don’t move, so old homes don’t free up for the young — new family formation requires new construction, sustaining housing demand even in a shrinking population.
AI’s shock spreads across services rather than concentrating in regions (unlike manufacturing shocks), which may make the job losses easier to absorb and redeploy.
A counterintuitive AI claim: less inequality means higher rates. The rich save more and that saving depresses rates (Mian–Straub–Sufi); if AI compresses inequality, that saving glut shrinks and rates rise.
AI capex is itself a rate-raiser. ~$700bn of hyperscaler spend, mostly cash pulled from money-market funds, removes loanable funds and spends them — a double headwind.
The Volcker move is now self-defeating. At 100%+ debt-to-GDP, crushing inflation with high rates creates a deficit so large you need future inflation to dissolve it.
Brazil is the cautionary tale. Its new spending is tiny and rule-compliant, but interest on debt drives an 8–9% deficit — proof that an early policy mistake can lock in punishing rates.
“Unanchored” = the central bank can no longer single-mindedly target inflation. When inflation and debt stability collide, it protects the bond market and lets inflation slip.
Change won’t be smooth. The structural shift in the central-bank/government relationship likely needs a crisis to force it; the obvious fixes are politically near-impossible.

Claude’s Take

This is the good version of a macro interview: a guest with a genuine thesis, a host who actually read the book and pushes back, and a clear chain of logic from demography all the way to “your central bank will pick the bond market over price stability.” The argument is coherent and the China-as-disinflation-engine framing is the cleanest I’ve heard it — once you accept that goods prices were exported from China, the claim that disinflation was borrowed rather than earned does a lot of work.

Two things to keep honest about. First, Pradhan’s track record cuts both ways. The Great Demographic Reversal called the inflation comeback before it was cool, which earns him real credit — but the 2021–23 inflation spike had a lot to do with pandemic supply chains and stimulus, not just demography, and it’s easy to retrofit a structural story onto a cyclical event. Higher inflation “for structural reasons” has been predicted for a long time by people who were early-and-wrong as often as early-and-right. Second, he’s admirably disciplined about uncertainty — he repeatedly separates “what we know” (demography) from “what might offset it” (AI, fixes) and refuses to lean on hope. That intellectual honesty is exactly what makes the bearish-on-central-banks conclusion more persuasive, not less.

The weakest link is the AI section, which he’d be the first to admit. “AI reduces inequality, which raises rates” is a clever two-banks-shot, but it rests on ignoring the top 0.2% and on one Jackson Hole paper; it’s a hypothesis dressed as a chain. Still, even granting all the caveats, the central mechanism — that a heavily indebted government structurally compromises its central bank’s inflation mandate — is hard to argue with and under-discussed. An 8: thoughtful, original, and the kind of framework that changes how you read every “will they cut?” headline. Not a 9 because the empirical case is still more elegant theory than settled fact.

Central Banks Are Losing Control | Ideas Lab | Ep.50