Potential: the speed limit for the economy

Every few weeks someone describes potential growth as the economy’s speed limit. The metaphor is useful, but it is attached to the wrong thing. The real speed limit is potential output, the level: the amount the economy can produce sustainably with inflation at target. Potential growth is only the rate at which that limit itself moves. The two answer different questions, and mixing them up produces bad readings of the cycle.

This post uses a small model of Australian potential output to make one point. Conventionally, the output gap is the difference between actual and potential output, a difference between two levels. This model goes one step further and splits that observed difference into an inflation-related component and a residual. Potential growth tells you how fast the potential level is moving. It does not tell you the sign of the gap, and neither potential growth nor observed GDP growth on its own tells you where inflationary pressure sits.

Two levels and the distance between them

The chart shows the two levels. Potential is the smooth dashed line. GDP wanders around it. In 2020 GDP fell about seven per cent below the line and potential barely moved. Through 2022 and 2023 GDP sat well above the line. In the June quarter of 2026 it sits just a little above it.

The tempting reading is that the distance between them is the output gap, but it is not. The model splits output three ways: potential, a gap read off inflation, and a residual for whatever neither accounts for. In the June quarter of 2026, GDP sits 0.28 per cent above potential while the gap, from inflation running 1.1 points above target, is 0.51 per cent. The residual makes up the difference at minus 0.23, and it is not a rounding error: its standard deviation is close to 1 per cent of GDP against 0.47 for the gap, so over the sample as a whole most of what you can see on this chart is residual rather than gap. The next chart shows the gap alone.

The inflation-defined gap was negative from about 2015 to 2021, sitting near minus 0.4 per cent. It turned sharply positive in 2022, peaked near 2 per cent in early 2023, and closed almost completely by 2025. Over the past few quarters it has opened again. Trimmed mean inflation bottomed in early 2025 and has been rising since, and because the gap is read off inflation, rising inflationary pressure and a widening gap are the same observation. The model puts the June quarter 2026 gap at plus 0.5 per cent, with a 90 per cent band of 0.3 to 0.7.

I should say plainly what this chart is. In this model the gap is defined as a constant times inflation’s distance from the 2.5 per cent target. The gap chart is therefore the trimmed mean inflation deviation with a rescaled axis. The path it traces, negative through the undershoot years, positive through the post-pandemic overshoot, is the path of inflation. That is the model’s central claim rather than a weakness: this is what the textbook definition of potential output looks like when you take it literally. Potential is the level of output at which inflation sits at target, so the gap is wherever inflation says it is.

Two things follow from the level view. First, the gap is a level concept, not a growth rate. A slowing or quickening potential growth rate says nothing on its own about whether the economy is above or below capacity. Second, actual growth relative to potential growth tells you what happened to GDP relative to the capacity line, but it does not by itself tell you what happened to the inflation-defined gap. In this model the residual moves too, and it can move enough to break the link between the two.

What potential growth tells you

Potential growth has fallen by about two percentage points since the late 1990s: 4.1 per cent in 1997, 3.1 in 2005, 2.8 in 2012, 1.9 in 2019. It dipped to about 1.8 per cent in 2020 and has recovered to 2.1, where it has sat since 2022. The credible band widens at the end of the sample, as it should. The model has less to go on for the most recent quarters.

None of this tells you the gap. A slowing potential growth rate does not mean the economy is above or below capacity. It means the level of capacity is rising more slowly than it was. Where the economy sits relative to that slower-moving line is a separate question, and in this model inflation answers it.

Potential growth does tell you the rate at which the capacity line is moving. With potential growth at 2.1 per cent, GDP growth above 2.1 per cent raises GDP relative to that line and growth below it lowers GDP relative to it. But observed GDP also contains the residual, so that comparison does not mechanically tell you whether the inflation-defined gap has opened or closed. Growth at potential is not the same as being at potential, and growth below potential is not the same as a positive gap closing. The number on its own is not enough. You need the level too.

June 2026 illustrates why the two should not be collapsed. Actual growth is about 2.2 per cent and potential growth about 2.1 per cent, almost identical, yet the inflation-defined gap has widened to about plus 0.5 per cent as inflation has drifted up over several quarters. The residual reconciles the two. Growth relative to potential tells you what happened to GDP relative to the capacity trend. Inflation tells this model what happened to inflationary pressure. They are related, but they are not the same observation.

How the estimate is built

The model has four lines and two data series: log real GDP from ABS 5206.0 and annual trimmed mean inflation from ABS 6401.0, quarterly from 1993.

y*_t      = y*_{t-1} + g_{t-1} + e_y     potential: a random walk with drift
g_t       = g_{t-1} + e_g                drift: itself a random walk
gap_t     = c (pi_t - 2.5)               gap: inflation's distance from target, rescaled
log_gdp_t = y*_t + gap_t + e_c           output: potential plus gap plus a residual

There is no Phillips curve. A central bank that succeeds in holding inflation near target removes the covariance between activity and inflation that a Phillips curve needs, so a slope estimated on the inflation-targeting era is biased toward zero (McLeay and Tenreyro, 2019). What survives policy is the definition of potential itself, and the model uses that instead of a slope.

The parameter c converts percentage points of excess inflation into per cent of output. It is estimated, but not by regressing inflation on anything. Given the trend and the inflation-defined gap, c is whatever best reconciles the two with observed GDP. It comes out at 0.47, with a 94 per cent credible interval of 0.26 to 0.69, clear of zero. One point of excess inflation implies about half a per cent of output gap.

The fourth line matters more than it looks. An earlier version defined potential outright as GDP less the inflation gap, with no residual. That version reproduced GDP exactly, so every movement in output that inflation did not explain was forced into potential. Potential fell seven per cent in the June quarter of 2020 and recovered within a year, which is not a description of productive capacity. Adding the residual fixes it. The pandemic period now sits in e_c, unexplained, which is where it belongs.

The chart of actual growth against potential growth needs one caution. The distance between the two lines is not the change in the inflation-defined gap. In this model it is the change in the gap plus the change in the residual, and the residual accounts for much of the short-run variation. The 2020 wedge is almost entirely residual: the model puts the gap that quarter at minus 0.6 per cent. Read the dashed line as the rate at which the capacity line is moving, and treat the shaded areas as a rough guide rather than a measure of the gap.

What the model cannot say

Measured in variance terms, the inflation-defined gap accounts for roughly a fifth of GDP’s deviation from potential. The gap has a standard deviation of 0.47 per cent of GDP, against 1.05 for the total deviation, and 0.47 squared over 1.05 squared is about 20 per cent. That is not a failure. It is the answer to the question the model was built to ask: how much of the cycle does inflation identify on its own? About a fifth. The remainder is variation the model does not attribute to the inflation-defined gap, and it leaves that variation in the residual rather than forcing it into potential.

c is probably too small. Phillips curve estimates from settings where policy does not suppress the relationship imply a value nearer 2.5 or 3, and the difference is the attenuation described above. Nothing inside this sample recovers it. The estimated gaps are therefore probably compressed toward zero. Their sign and timing are more informative than their magnitude. Everything is also conditional on the 2.5 per cent anchor. A shift in the target, or in what people believe the target to be, would show up as a gap.

The level of potential is the object of interest. Its growth rate tells you how fast the capacity line is moving, and inflation tells you where the economy sits relative to it. Neither potential growth nor observed GDP growth on its own tells you the inflationary gap. Hold the level and the rate apart and the charts read cleanly. Collapse them into one number and they do not.