When a central bank moves the cash rate, it’s acting on a specific model of how the economy works. That model is the New Keynesian synthesis, and it’s the closest thing modern economics has to an operating manual for monetary policy. Every inflation target, every rate decision, every line in a central bank statement about “returning inflation to target” comes out of this framework. Understanding it is the difference between watching the news cycle of rate rises and cuts and actually seeing the logic behind it.
This isn’t a history of how these ideas came together, nor an account of what exactly got synthesised between classical and Keynesian economics. It’s a walk through the intuition: how the pieces of the model fit together, and where they hold up worse than the theory suggests.
Output and Output Gaps
Every economy has a speed limit. Economists call it potential output, or Y*. It represents how much an economy can produce when its workers, factories, and capital are all fully but sustainably employed. Actual output (also known as GDP or Y) moves around this limit. The difference between the two, Y minus Y*, is called the output gap. Sometimes it’s negative and Y sits below Y*, and resources go idle. Sometimes it’s positive and Y pushes above Y*, and something has to give.
That “something” is usually price. When Y runs above Y*, firms want more workers and more raw materials than the economy can comfortably supply. They compete for those scarce inputs by paying more for them. Wages rise faster than productivity. Input costs climb. Firms pass these costs on, and the general price level starts to rise. This is inflation, denoted π. In practice, "inflation" isn't one number. Statisticians and central banks track it several different ways; headline CPI, trimmed mean, weighted median, wage growth, and more, and while they mostly move together, they can diverge for stretches at a time.
Potential output also has a speed, usually called g*: how fast the economy can grow without producing inflation, the pace at which Y and Y* can rise together and stay aligned. If GDP grows faster than g* for long enough, it isn’t just a one-off gap that opens up, growth itself is outrunning the economy’s capacity, and the pressure described above builds year after year rather than once.
From Output Gaps to Unemployment
That story works cleanly in theory, but it assumes we can actually see where Y and Y* sit. In practice, we can’t.
Y* is hard to observe in real time. Nobody can measure “potential output” the way they measure GDP. Economists translate the output gap into something with a more direct, observable analogue: the labor market.
When firms can’t get enough workers, unemployment falls below some baseline rate. Economists call this baseline the NAIRU, the non-accelerating inflation rate of unemployment, or more simply u*. When actual unemployment (u) drops below u*, workers gain bargaining power. Wage growth accelerates, and wage growth feeds into price growth.
This brings us to the Phillips curve. In its simplest form, it says inflation rises when unemployment falls below u*, and falls when unemployment rises above it. A tight labor market is expensive labor, and expensive labor becomes expensive output.
Demand pushes output above potential. A tight labor market follows, with unemployment below u*. Wages rise faster than the economy can absorb. Prices rise. Unemployment isn’t just a welfare statistic. It’s a leading indicator of price pressure.
Why Expectations Matter So Much
The original Phillips curve broke down in the 1970s. Unemployment and inflation rose together, something the simple relationship said shouldn’t happen. The fix was to recognize that the relationship isn’t really between unemployment and inflation. It’s between unemployment and unexpected inflation.
If everyone already expects prices to rise 2.5% next year, that 2.5% gets built into contracts, wage negotiations, and price-setting decisions before a single extra job is filled. Actual inflation ends up being roughly what people expected, plus or minus a bit depending on where unemployment sits relative to u*. Expectations aren’t a footnote. They’re the baseline everything else gets added to.
This isn’t just about extrapolating the recent past. Firms setting prices and workers negotiating wages are also pricing in where they expect inflation to be by the time that price or wage next gets revisited, which means expectations pull in information about the future, not only memory of what’s just happened.
If expectations are anchored near a stated target, say 2.5%, the central bank only has to manage the gap around that target. Most of the heavy lifting is already done by public belief.
The danger comes when expectations become unanchored. If people start expecting 5% inflation because that’s what has actually happened for a while, that 5% becomes the new baseline. The central bank is no longer managing a temporary demand gap. It’s fighting a self-fulfilling expectations spiral.
A central bank that convincingly commits to a target doesn’t just react to inflation. It shapes the expectations that determine inflation in the first place.
This is also what anchored expectations buy the central bank on the way in, not just the way out. Not all inflation is the same kind of problem. A supply shock, an oil spike, a shipping disruption, pushes prices up for a while regardless of how tight the labor market is. If expectations are well anchored, the central bank has more room to look through a temporary supply shock. It does not necessarily have to force inflation immediately back to target by crushing demand. But the longer or broader the shock lasts, the harder that judgement becomes: potential output may itself have fallen, and keeping demand unchanged can leave the economy running above its newly reduced capacity.
That protection isn’t automatic, though. It depends on wages actually being set with reference to expectations rather than mechanically. Some wage-setting is formally indexed to inflation instead: cost-of-living clauses, automatic public sector adjustments, some sectoral bargaining agreements. In those cases, the link from prices to wages doesn’t run through anyone’s beliefs at all. It’s contractual. A shock pushes prices up, wages rise to match by rule, and those higher costs push prices up again, regardless of how credible the central bank is or how well-anchored expectations otherwise are. Anchoring expectations is much less effective against a mechanical price spiral.
The “New Keynesian” part of this story rests on a simple friction: prices and wages don’t adjust instantly. Firms reset prices periodically, contracts last, wages get negotiated at intervals, and businesses are reluctant to move prices continuously ahead of their competitors. That stickiness is what gives the central bank something to work with. A change in the nominal interest rate briefly changes the real interest rate too, since prices haven’t caught up yet, and spending responds before every price in the economy has had time to adjust. Without that friction, the cash rate would have far less power over real activity in the short run.
The Cash Rate and How It Works
A central bank can’t directly control output, unemployment, or inflation expectations. What it controls is the price of very short-term money, the cash rate.
The theoretical version of this works through what’s called the IS curve, a name that’s largely an artefact of history. When the real interest rate sits above r*, its neutral level, households and firms postpone spending and investment, and output falls below potential. When it sits below r*, demand is stimulated and output rises above potential. That’s the clean version: the interest rate gap drives the output gap.
The IS curve takes all of that complexity and squashes it into one line: rates above neutral cool the economy, rates below neutral heat it up. Reality is messier. The effect runs through mortgages, business borrowing, asset prices, the exchange rate and bank lending, and the mix changes from country to country. Attempts to estimate the clean version directly often find a flat or wrong-signed connection between the real rate gap and output, and in its textbook forward-looking form the equation can also imply implausibly large effects from interest-rate promises far into the future, the well-known “forward guidance puzzle.” Real households and firms appear much less responsive to distant policy promises than the frictionless logic implies. It’s the channels themselves, not the clean equation, that do the work below.
Raising the cash rate works mechanically first. Banks price loans and deposits off the cash rate, so mortgage rates, business borrowing costs, and savings rates all move with it. Longer-term rates, the ones that matter for a thirty-year mortgage or a company’s investment plans, move too, since they largely reflect the expected path of future short-term rates.
Higher borrowing costs discourage business investment and household spending on big-ticket items like cars and houses. Existing variable-rate borrowers face higher repayments, which squeezes disposable income directly. Asset prices tend to soften too, since a higher discount rate lowers the present value of future earnings, creating a wealth effect that further dampens spending. In open economies, higher rates attract foreign capital, the currency appreciates, imports become cheaper, and exports become less competitive, both of which weigh on demand.
It’s tempting to picture this working mainly through mortgage repayments, since that’s the channel households feel most directly. In Australia, though, the RBA’s own modelling suggests the exchange rate channel carries more of the weight, especially for inflation, while the aggregate cash-flow effect is smaller than the household-level story implies.
The two channels have different sources. Australia’s open economy and floating currency make the exchange rate channel unusually important. Its high share of variable-rate mortgages makes household cash flow respond unusually quickly, but that speed doesn’t translate into the largest aggregate effect. The mix isn’t universal either. In an economy with mostly fixed-rate mortgages, like the United States, the cash-flow channel works on a much longer lag, and the balance between channels looks different again.
A rate rise also sends a signal. It tells households and firms the central bank is serious about controlling inflation, which itself helps keep expectations anchored. A credible central bank can shift behavior just by moving the rate slightly, or even by signaling a future move through forward guidance, because people trust the target will be met. A less credible bank has to move rates much further to achieve the same effect.
This activity, mechanical and psychological, brings demand back toward potential output, easing the labor market and taking pressure off wages and prices. But it takes time. Mortgage resets happen unevenly. Investment decisions take months or years to unwind. A common rule of thumb is twelve to eighteen months, sometimes longer, for a rate change to fully show up in inflation data.
None of this is permanent. A rate change doesn’t relocate the economy to some new setting and leave it there. Its effects fade as the economy adjusts, which means the central bank isn’t solving a problem once, it’s re-solving it continuously, watching where output, unemployment, and inflation sit relative to potential and adjusting again as conditions shift.
The Trouble With the Star Variables
The central bank is steering by three numbers it cannot actually observe:
- Y*, the potential output for the economy,
- u*, the non-accelerating inflation rate of unemployment (or NAIRU), and
- r*, the neutral real interest rate, the rate consistent with stable inflation.
The central bank’s own actions distort the data it relies on. Estimates of r* are built from the observed relationship between growth, inflation, and real rates, and a decade of quantitative easing changed that relationship by deliberately compressing yields and financial conditions. There’s no clean way to tell how much of the resulting drop in measured r* reflected a genuine shift in the economy’s equilibrium and how much reflected a policy regime that had itself changed the relationships from which r* was being estimated.
Before the financial crisis, central banks were genuinely guided by something close to a Taylor rule, built on a simple insight: to actually tighten policy, the rate has to rise by more than the increase in inflation, not just match it. Even then the rule was getting ragged. The global saving glut, a wave of savings out of China that pushed down interest rates worldwide, was already pulling the neutral rate away from where the models assumed it sat. After the crisis, the rule broke down as an operating guide entirely. The mechanism behind it, that inflation above target demands more than a one-for-one response, never went away. What went away was the willingness to mechanically follow a formula built on a neutral rate nobody could pin down anymore.
Still the Working Framework
None of the individual stars are pinned down precisely. But they aren’t estimated in isolation, either. A tight labor market showing up alongside output running hot, a falling r* coinciding with weak investment demand, these are different pieces of data telling a consistent story. That corroboration across variables is itself a kind of evidence. The model doesn’t need any single star to be exact to be doing real work, it needs the overall picture to keep holding together, and mostly it does.
That doesn’t make the estimation problem go away. The stars are still poorly measured, central bank actions still pollute the data used to measure them, and the 2021-23 inflation surge along with the QE-to-QT transition have left a genuine mess to sort through. But no rival framework has displaced the model despite all of that. No major central bank runs policy on strict money-supply targeting, or nominal GDP targeting, or anything else at scale. What every serious central bank carries instead is some version of this model in its head: a sustainable level of activity, inflation that responds to how far conditions sit from that level, a policy rate that gets tightened or loosened to close the gap. The formal Taylor rule fell out of use. The underlying picture didn’t.
The New Keynesian synthesis survives less because it’s been vindicated by events and more because nothing better has come along to replace it.
No comments:
Post a Comment