Inflation Targeting vs nGDP Targeting

What Are We Talking About

Today's blog post is very technical. We are talking about which variable should be the focus of a central bank when it sets interest rate policy. Most of the world's central banks use an inflation target, typically around 2 per cent, although Australia has a 2 to 3 per cent target band and within that it targets the 2.5 per cent mid point. An alternative which is often promoted is nominal gross domestic product (nGDP) targeting, where the bank aims to maintain a steady growth rate or path for nGDP.

The Shared Foundation

Both camps share more than they disagree on. Both accept that excessive or volatile inflation hurts workers and businesses alike, by eroding real wages, damaging investment, and making the future harder to read. Both accept that recessions cause real damage and that the central bank can do useful work stabilising the business cycle. Both accept that a central bank with discretion over short-term interest rates should be independent, operate under a clear mandate, and anchor expectations as one of its main jobs.

The disagreement is narrower than the public debate sometimes makes it look. It is about which nominal aggregate the central bank should defend. Inflation targeting says defend the price level, with a target inflation rate of around 2 to 3 per cent. nGDP targeting says defend the path or growth rate of total nominal spending, with a target that combines inflation and real growth into one number. Both are flexible frameworks that allow for central bank discretion. Neither is a mechanical rule.

The argument is about which variable provides the better anchor, and what "better" even means. Better at delivering stable prices? Better at protecting employment? Better at preventing expectations from coming loose? Better at handling supply shocks without overreacting? The two frameworks deliver different answers to these questions, and the disagreement is partly about which question matters most. That is a value choice as much as a technical one.

What nGDP Targeting Actually Is

Nominal gross domestic product is the dollar value of everything the economy produces in a given period, before adjusting for inflation. Real GDP growth plus inflation equals nGDP growth. (A quick side note: When economists use the acronym GDP on its own, they almost always mean real GDP, and they make it clear when they are talking about nominal GDP). The framework asks the central bank to target the sum rather than either component separately. A typical proposal might aim for 5 per cent nGDP growth annually. If real growth comes in at 2.5 per cent, inflation runs at 2.5 per cent. If real growth slows to 1 per cent, inflation is allowed to drift up to 4 per cent to fill the gap. The central bank stays on target either way.

The more popular version is level or path targeting rather than growth-rate targeting. Under growth-rate targeting, the central bank aims for a steady nominal growth rate each year, say 5 per cent, and judges each year on its own. If nGDP grows at 3 per cent during a recession, the missed output is gone and policy returns to aiming for 5 per cent going forward. Under level targeting, the central bank commits to a specific path for the dollar value of nGDP. If the economy undershoots during a recession, policy stays loose afterward to catch the level back up. The catch-up can run for years, not just a single annual reset. This is the feature advocates value most. The commitment to reverse past shortfalls is supposed to anchor expectations in a way growth-rate targeting cannot, because households and firms know any demand collapse will be made good rather than tolerated. The recovery from a downturn is supposed to be hotter than normal to compensate for the lost output.

The per capita refinement strips out population growth. Total nGDP can rise simply because the population is growing, which tells you nothing about whether people are getting better off. Per capita nGDP targets the dollar value of output per person, which more closely tracks what affects household welfare. The framework also handles demographic change more gracefully. A country with falling labour force growth automatically gets a lower implied nominal target without anyone needing to revise the headline number. Per capita targeting has gained ground among advocates like David Beckworth precisely because it adapts to the demographic transitions that fixed total nominal targets struggle with.

Both versions ultimately reduce to the same equation. The target equals desired inflation plus assumed trend real growth, with the per capita version using per-worker productivity instead of aggregate output. Either way, the central bank is committing to a nominal aggregate that bundles inflation and real growth into a single number.

The Case for nGDP Targeting

The intellectual case is real and worth stating fairly. nGDP targeting has several genuine advantages over inflation targeting in theory.

The first is that it handles supply shocks more cleanly. When oil prices spike or a pandemic disrupts shipping, headline inflation rises even though the underlying problem is reduced supply rather than excessive demand. A pure inflation targeter tightens in response, compounding the supply shock by squeezing demand on top of it. nGDP targeting looks past the supply-driven inflation and responds only when nominal spending itself is off track. The 2008 episode is the canonical example. Oil prices pushed headline inflation above 5 per cent in the US while nominal demand was already collapsing. The Fed held policy too tight, and the framework provided no natural diagnostic for the mistake.

The second is that nGDP captures what matters for household balance sheets. If the central bank keeps total nominal spending growing steadily, the wage bill the economy can support also grows steadily, which means firms do not have to fire workers or cut nominal wages during downturns. Households servicing nominal debts on mortgages and other loans benefit from steady nominal income growth. The framework targets the variable that most directly protects employment and wage stability, rather than the variable that affects the price index. Inflation targeting, by focusing on prices, can let nominal incomes collapse during a recession while still hitting its target, which is exactly the kind of episode that produces unemployment spikes and debt crises.

The third is that level targeting provides a stronger commitment device at the zero bound on policy rates. If nGDP falls below path, the central bank commits to overshooting later to catch up. This changes expectations during a downturn in ways that inflation targeting cannot match. The framework borrows from Krugman and Woodford's work on managing expectations when policy rates cannot fall further.

Underneath the technical arguments sits a more honest version of the case that advocates rarely state directly. nGDP targeting reweights the central bank's mandate toward employment. Most modern central banks have an employment and inflation objective in some form, but in practice they treat inflation as the more important constraint and let employment outcomes fall out. nGDP targeting reverses this priority by making nominal income stability the headline goal, which means accepting more inflation variability in exchange for less employment variability. Advocates think this trade is desirable because the welfare cost of employment fluctuations is larger than the welfare cost of modest inflation variability, and because the framework institutionally commits the central bank to a more employment-friendly stance than its inflation-targeting instincts would deliver. The political coalition supporting the framework, particularly on the centre-left during the 2010s, was attracted to this reweighting more than to the technical apparatus. This is the framework's strongest claim and also where its sharpest disagreement with inflation targeting lies.

These are not silly arguments. The 2008 to 2012 period exposed real weaknesses in inflation targeting, and the intellectual case for nGDP targeting has done well in academic discussion for good reasons.

The Case Against

The case against is partly that the advantages above are overstated, and partly that the framework introduces problems of its own that the advocates rarely acknowledge.

Start with the supply shock claim. Inflation targeting is not actually mechanical. Every serious central bank already distinguishes between demand-driven and supply-driven inflation, and weighs both against what is happening to expectations. Staff produce decompositions at every meeting. The framework permits judgement about whether to look through supply shocks, conditional on expectations staying anchored. The 2021 to 2023 episode showed this in real time. Central banks initially diagnosed the inflation surge as supply-driven and held policy accommodative. As wage growth accelerated, inflation spread across categories, and survey measures of expectations started drifting upward, they updated and tightened. Whether they updated quickly enough is a fair debate. The point is that the framework allowed adaptation once the persistence became evident, rather than locking the central bank into a fixed response.

The expectations channel is the deeper point. Central banks do not just ask whether inflation is demand-driven or supply-driven. They ask whether the inflation is the kind that will pass through to wage and price setting behaviour and reset expectations at a higher level, or the kind that will fade as the shock dissipates. That judgement is built into how flexible inflation targeting actually operates. The framework gives the central bank room to look through a transient shock while keeping the inflation target as the public anchor that disciplines expectations. The whole architecture is designed around protecting the anchor, because once expectations move, getting them back requires a recession.

What inflation targeting refuses to do is formally encode the supply, demand and expectations decomposition into the policy rule. That is a feature, not a bug. The distinction requires judgement, and embedding the judgement in a formula does not improve it. nGDP targeting does not solve any of these problems. If oil prices spike, real growth falls, inflation rises, and nGDP stays on path, the framework's answer is to hold policy steady. The supply shock gets absorbed without policy overreaction, which advocates count as a feature. But the framework provides no signal about whether the inflation pass-through is the kind that will fade as the shock dissipates or the kind that will reset expectations at a higher level. The headline anchor has bundled inflation and real growth together precisely when you want to look at inflation in isolation to assess the expectations risk. You are back to using all the same diagnostic tools, with the added complication that the framework's headline number tells you nothing about the question that matters most.

The employment reweighting argument deserves direct engagement because it is the framework's most honest claim. The critique is not that the trade-off is illegitimate. Central banks do face a real choice about how heavily to weight employment versus inflation, and reasonable people can disagree about the right weights. The critique is that the trade-off the framework promises is more uncertain than its advocates admit. The argument for accepting higher inflation tolerance rests on getting employment gains in return. The post 2008 experience suggests this exchange is not reliable. When the Phillips curve is flat, as it was through the 2010s in both the US and Australia, running the economy hot looks low cost because inflation barely rises. But the same flatness means looser policy delivers little employment benefit either. The framework ends up accepting more inflation risk without cashing in the employment gains the trade was supposed to buy. Worse, the curve can steepen quickly once labour markets tighten and supply gets constrained, as the post pandemic experience showed. A framework that institutionalises greater inflation tolerance looks safe during the flat phase and becomes dangerous exactly when the curve comes back. The problem is not that employment should receive no weight. The problem is that the exchange rate between inflation risk and employment gains is too unstable to be hard-coded into the headline target.

The Productivity Problem

The deeper objection to nGDP targeting is the one its advocates rarely engage with seriously. The framework requires you to know something you cannot know in real time.

An nGDP target combines an inflation goal with an assumed trend real growth rate. To set the target you must pick a number for trend productivity. If productivity slows and you fail to update the target, the equation forces the residual into inflation. Lower real growth plus the same nominal target equals higher inflation. There is no third option. The central bank cannot conjure productivity by printing money.

An informed advocate will object that this misrepresents the framework. The nominal target is supposed to be primitive rather than derived. The central bank simply picks a stable nominal spending path and lets inflation and real growth divide the total however the supply side dictates. Productivity is not an input to the framework. It is whatever the economy produces given the nominal anchor.

This response is formally correct but incomplete. In practice no nominal target is politically or economically neutral. A 5 per cent nominal target implicitly assumes society is willing to tolerate whatever inflation rate results from the economy's actual productivity performance. If productivity growth collapses from 2.5 per cent to 0.3 per cent, the same nominal target now implies inflation near 4.7 per cent rather than 2.5 per cent. The productivity assumption is not absent from the framework. It is embedded implicitly in the choice of nominal path rather than stated explicitly. Pretending the target is primitive does not make the assumption go away. It just hides it inside the chosen number where it cannot be debated openly.

The Australian data makes the problem concrete. Note, this charts uses the logarithm of nGDP so that fixed growth rates appear as straight lines. 

The chart shows the log level of Australian nGDP per capita against two candidate target paths, both anchored at the start of 1993. The red dashed line is a flat 4.2 per cent annual path, which embeds the 2.5 per cent inflation target plus 1.7 per cent assumed real per capita growth. The green dashed line is a kinked path running at 5.25 per cent before the GFC and 3.25 per cent after, which embeds the same inflation target but allows real per capita productivity to step down from 2.75 per cent to 0.75 per cent around 2008.

Two things jump out. The flat 4.2 per cent path runs below the actual data for almost the entire sample. The gap opens up through the 1990s as productivity ran above the embedded assumption, sits widest from the early 2000s through to the late 2010s, then narrows sharply post-pandemic as inflation pushed nominal growth higher than the path. Under strict level targeting, the central bank would be looking at this chart and concluding that policy has been persistently too loose for most of three decades. That conclusion is absurd. What it actually shows is that the assumed productivity rate baked into the 4.2 per cent target was wrong. The 1990s and 2000s could deliver more than 4.2 per cent nominal growth per capita without inflation problems, because productivity was higher then.

The kinked green path fits the actual data almost perfectly through to the pandemic. This is what level targeting would have needed to look like to be honest. The framework would have needed to recognise around 2010 to 2014 that productivity had structurally slowed, and revise the target path downward by two percentage points per year. The question becomes whether any framework can credibly do this.

Productivity Is Less Stable Than The Framework Requires

The productivity numbers underneath the kinked path tell their own story.

The chart shows Australian labour productivity growth with a 10 year moving average that smooths out the quarterly noise. The trend climbed from around 1 per cent in the late 1980s to a peak near 2.4 per cent around 2002, slowly fell back through the late 2000s, sat around 1.0 to 1.3 per cent through the 2010s, and has dropped to 0.3 per cent over the past few years. Productivity is at its lowest level in the entire 35 year window the chart covers.

This is the framework's deeper problem. Productivity does not deliver a stable number that a central bank can confidently embed in a nominal target. The smoothed trend moves around by more than 2 percentage points across the sample. The peak in the early 2000s now looks like an episode rather than a regime. Anyone trying to identify a structural trend would have changed their mind several times. The late 1990s and early 2000s would have looked like a permanent productivity upgrade from the Hawke and Keating reforms plus the IT boom. The 2010s would have raised questions about a new lower normal. The current numbers would have confirmed those questions decisively. Each updated view was reasonable given the data at the time, and each was wrong about what came next.

The kinked path in the level chart needed another downward kink that did not appear because the chart stops at 2025. Honest level targeting now needs to assume something like 0.3 per cent productivity growth, not the 0.75 per cent embedded in the green path. The current decade target should be around 2.8 per cent nominal per capita growth, not 3.25 per cent. The framework requires yet another path revision.

And that is assuming the current 0.3 per cent reading is the right number to use, which is itself unknowable. The current weakness could be a structural shift that will persist or worsen as ageing, declining business dynamism, and energy transition costs bite further. It could also be a cyclical trough that will partly reverse as the post pandemic disruptions fade and AI adoption starts to show up in measured productivity. The honest answer is that we will not know for another decade. A central bank setting a nominal target today has to commit to one view or the other. If it assumes 0.3 per cent productivity and the actual delivery turns out to be 1.0 per cent, the nominal target will be too low and the framework will deliver insufficient nominal growth, which means inflation undershooting. If it assumes 1.0 per cent and actual productivity stays at 0.3 per cent, the target will be too high and the framework will deliver excess inflation. Either error compounds over the framework's commitment horizon.

This is the inescapable problem with embedding a productivity assumption in a headline target. The assumption is needed because the framework defines the target as inflation plus real growth, and real growth depends on productivity. The assumption cannot be made well because productivity is neither stable nor predictable. Inflation targeting sidesteps the problem by not requiring the assumption. Productivity surprises still affect the inflation targeting central bank through their effects on NAIRU, the natural rate of interest, and fiscal sustainability. But the framework does not mechanically translate productivity surprises into target drift. Whatever the economy delivers in real growth, the framework absorbs it as real growth rather than as inflation overshoot.

The Real Time Problem

Defenders of nGDP targeting respond that the target should be updated as productivity estimates change. This is reasonable in principle but defeats much of the point. The framework is supposed to provide commitment. If targets get revised every few years based on rolling productivity estimates, the commitment evaporates and you have discretionary policy with extra steps. The kinked path requires exactly the kind of revision the framework's theoretical case says it should not allow.

The deeper problem is that productivity slumps are diagnosed years after they begin. Cyclical and structural productivity components both move slowly and overlap, which makes them genuinely hard to separate in real time. By the time you have enough evidence to confidently distinguish a structural shift from a cyclical wobble, the trend break is ancient history. A central bank operating under nGDP targeting through the 2010s would have spent the entire decade with inflation drifting above target before the framework could be updated. The level gap would compound year after year, with the central bank running loose policy to chase a path the economy could no longer sustainably reach. Inflation would run hot because the framework is pushing for nominal growth the supply side cannot deliver. Real growth would stay weak because the supply side is what it is. The catch-up would never happen. That is a multi-year recipe for stagflation. Expectations would come unanchored along the way. The eventual reset would require a recession, exactly what inflation targeting is designed to avoid.

The post pandemic period adds a further wrinkle. The actual data now sits well above the kinked path, with the overshoot driven mostly by inflation rather than real growth. Under strict level targeting, the central bank should now be committed to engineering below trend nominal growth for several years to bring the level back to path. No central bank would deliberately do this. So the framework's catch up logic only works in the direction of accommodating shortfalls, not in the direction of reversing overshoots. The symmetric commitment exists only in theory.

Inflation targeting handles all of this better because it does not embed a productivity estimate in the headline target. When productivity disappoints, real growth slows and inflation stays roughly at target. The economy gets poorer in real terms, which is the honest signal. The framework forces the productivity question into the open rather than hiding it inside the monetary aggregate. The 2.5 per cent inflation target does not require the central bank to pre-commit to either productivity number. That robustness to productivity surprises is the framework's deepest virtue.

Expectations Are The Asymmetric Risk

The productivity argument matters because of what it does to expectations. Inflation is very hard to control once expectations start to wander. The 1970s established this and central banks have spent the four decades since building institutional commitments designed to prevent expectations drift from starting in the first place.

The mechanism is familiar but worth restating. Once households and firms start expecting higher inflation, they incorporate it into wage demands, price setting, and contract design. Each piece of the economy starts reasoning about inflation rather than around it. The shift is not gradual. It tips. Once it tips, only deliberate engineered slack can reverse it, which means recession. The Volcker disinflation in the US, the equivalent moves across Europe, and Australia's own 1990 to 1991 recession all illustrate the cost of letting expectations come unanchored. The cost was measured in years of high unemployment and lost output.

The Australian experience is particularly instructive because the country took the slow path. Fraser talked tough about inflation in the late 1970s but did very little to deliver it. The Hawke government deliberately chose gradualism through the Accord, trading wage restraint for social wage concessions and avoiding the orthodox Volcker-style disinflation. The strategy worked partially. Inflation drifted down through the 1980s but never anchored. The 1988 to 1989 boom pushed it back up. By 1990 the country was running 8 per cent inflation and had to take the recession it had been postponing for a decade. Australia was roughly ten years late to the disinflation party, and the eventual reset was probably more painful than it would have been in 1982.

This is what nGDP targeting risks. The framework can produce extended periods of inflation running above any sensible target, with the central bank reporting that nominal growth is on path. A decade of 4 per cent inflation paired with 1 per cent real growth would hit a 5 per cent nominal target perfectly. The framework would call this success. Expectations would not. Households and firms would observe a decade of inflation running well above the implicit price stability norm and adjust their behaviour accordingly. By the time the central bank or its political masters decided the inflation experience was unacceptable, the expectations damage would already be done and the reset would require the kind of recession that institutional commitments are supposed to prevent.

The asymmetry of the cost structure makes this much worse. The more characteristic failure mode of inflation targeting is excessive cyclical weakness, where the central bank holds policy too tight and produces unnecessary unemployment until inflation falls back. The damage is real but cyclical and reversible. The more dangerous failure mode of nGDP targeting is validating persistent inflation when productivity disappoints. The framework would report success even as expectations come loose, and the eventual correction would require the kind of recession that institutional commitments are supposed to prevent. Frameworks should be judged by the worst case they can enable, not the average case. The frameworks are not symmetric in their dangerous failure modes.

There is a related point about institutional commitment. The inflation targeting framework's value comes partly from its refusal to permit politically convenient flexibility. The central bank cannot explain that higher inflation is acceptable this year because productivity is weak, or because real growth needs support, or because the nominal aggregate is below path. The framework forecloses those explanations. Foreclosing them is the point. nGDP targeting reopens exactly the kind of explanations that the inflation targeting framework was designed to prevent. The flexibility looks attractive in normal times. The historical experience suggests it becomes corrosive precisely when the institutional commitment is most needed.

The Australian Phillips Curve Has Returned

The case for nGDP targeting in the US debate was built during the 2010s, when the post-2008 flattening of the Phillips curve made traditional inflation targeting tools look less useful. Looser policy did not seem to produce inflation, which suggested the curve was broken and the central bank should focus on a different anchor. The Australian curve flattened too through the same period.

The post-pandemic experience has changed the picture in both countries. Recent US research shows that the slope steepened significantly between 2021 and 2024. The Australian curve has reasserted itself since 2021 as well. The post-COVID curve in both economies looks at least as steep as the pre-GFC versions. Wage growth responds to unemployment. Inflation responds to wage growth. The traditional transmission has come back.

The Australian chart shows the pattern clearly. The 1998 to 2002 and 2003 to 2008 curves slope downward as a Phillips curve should, with the second sitting at lower unemployment as the natural rate fell. The 2010 to 2019 curve is essentially flat. The post-2021 curve is sharply downward sloping again, at least as steep as the pre-GFC version. The relationship between unemployment and inflation died for a decade and came back.

This means the diagnostic apparatus underneath inflation targeting still functions. NAIRU estimates, while uncertain, cluster in a narrow enough band that policy can be set against them with reasonable confidence. The Phillips curve provides a cross check. Inflation surprises tell you whether your NAIRU estimate was too high or too low, and you can update. The whole New Keynesian apparatus of multiple star variables triangulating against each other is doing its job.

nGDP targeting would collapse this multi-dimensional system into a single variable. Instead of separate readings on inflation, unemployment, output, and rates, you get one number that bundles them together and a target path to compare it against. You lose the triangulation. When the framework misfires, there is no internal mechanism for catching the error.

The deeper point is about information. Aggregation reduces informational redundancy. A single sufficient statistic is elegant if the underlying model is correct and complete. It is dangerous if the model is incomplete, because the aggregate hides exactly the kind of inconsistencies that would reveal the model error. Inflation targeting keeps the components separate and lets the cross-checks do their work. If inflation, unemployment, wages, and the output gap tell mutually inconsistent stories, the central bank knows something is wrong and looks harder. Under nGDP targeting, the components are summed before they reach the policy rule. The mutual inconsistencies disappear into the aggregate. The framework looks clean precisely because it has thrown away the information that would tell you it is broken.

What About R-Star?

The natural rate of interest is the one star variable that genuinely is hard to estimate. I have written about this elsewhere. The Australian IS curve rate channel is weak relative to noise, which means r-star cannot be identified from standard models. The post-GFC era saw a large divergence between bond-market readings of r-star and trend-growth readings, and only partial reconvergence since. The framework debate has to acknowledge that this particular star is murky.

But the r-star problem does not favour nGDP targeting. If anything, it cuts the other way. nGDP targeting needs r-star to set policy just as much as inflation targeting does, because the central bank still uses interest rates as its instrument. The framework provides no escape from the r-star problem. It just adds a separate requirement to know trend real growth, which has the same kind of identification problems. You end up needing more uncertain inputs, not fewer.

The Honest Comparison

The fair statement is that both frameworks make demands on knowledge we do not have. Inflation targeting requires the central bank to estimate NAIRU, potential output, and the natural rate of interest. nGDP targeting requires all of the above plus an estimate of trend productivity growth embedded in the headline target. The framework with fewer embedded assumptions is more robust to being wrong about any one of them.

Inflation targeting also has the virtue of accumulated experience. The Taylor rule was not derived from first principles. It was a distillation of how good central banks actually behaved during a period when central banking was working. The around 2 per cent target emerged through trial and error across many countries. The framework absorbs lessons from failures and adjusts, as the 2020 Fed framework review demonstrated by incorporating the level-targeting insight from the nGDP literature without abandoning inflation as the headline variable.

nGDP targeting has never been tried by a major central bank. It is a hypothesis, not a practice. The challenger to a working framework carries the burden of proof, and that burden has not been met.

What This Means For Australia

The Australian inflation target of 2 to 3 per cent has been a reasonable framework choice. It has handled the 2010s period of below-target inflation, the post-pandemic surge, and the current disinflation in trimmed mean trends. The RBA can use the functioning Phillips curve to set policy with reasonable confidence. The 2 to 3 per cent band provides flexibility that a point target would lack, which has worked well in practice.

Nothing in the current situation calls for a framework switch. The intellectual contributions of the nGDP literature, particularly around the importance of nominal demand stability and the value of level targeting at the zero bound, have been partly absorbed into flexible inflation targeting. The headline framework can stay where it is. The diagnostic toolkit can quietly improve without changing the target.

The boring conclusion is that inflation targeting works well enough most of the time, fails sometimes in ways that can be partially fixed within the framework, and is more robust than its critics suggest to the kinds of mistakes it could make. nGDP targeting is more elegant on paper, more demanding of knowledge we do not have, and untested where it matters. Stick with what works.