Australian inflation took off in 1973 and did not get back below five per cent, and stay there, until 1991. Eighteen years. Germany peaked at 7.9 per cent and was under 5 within three. The United States peaked at 14.8 and took nine. We peaked at 17.5, and the Reserve Bank's own 1992 conference on disinflation ranked Australia poorly on the unemployment cost of getting there, putting it down to high real rigidities.
It was not for want of instruments. The Arbitration Commission set wages. Treasury set the budget. The exchange rate was managed. The Bank did what it was told. Every one of them moved the price level and not one of them was aiming at it. That is how you get eighteen years.
Then, briefly, we fixed it. Not by giving the job to the central bank. By giving every instrument a rule, and pointing the rules the same way.
That arrangement lasted about a decade. It has been coming apart ever since, bit by bit, one piece at a time, three or four years apart, and nobody has put the pieces on the same page.
What forty years taught us
Each of these was learned the expensive way.
Automatic propagation does not work. An automatic cost of living formula had adjusted the basic wage since the 1920s. In September 1953, after the Korean War wool boom took inflation to twenty five per cent, the Commonwealth Court abolished it, saying on the record that automatic adjustment had been an accelerating factor in rapid price increases. Twenty two years later its successor brought it back.
Because the second time confirmed it. Indexation returned in 1975 as a disinflationary measure, on the theory that orderly compensation would stop the catch-up claims. The Reserve Bank's own annual report records what those claims had produced: average weekly earnings up twenty eight per cent in the year to December 1974 against sixteen per cent inflation, which the Bank described as compressing three years of wage growth into twelve months. The formula may well have brought that down faster than the claims would have. It also wired the economy to transmit every subsequent shock, and locked in a real wage the country could not carry when the terms of trade turned.
Removing a mechanism without replacing it does not work either. The Commission terminated indexation on 31 July 1981. Bargaining went award by award. The Metal Industry settlement in December gave fitters twenty five dollars a week, another fourteen the following June, and cut standard hours from forty to thirty eight. Every other industry sought to match it. Five months after the system came out, there was a breakout.
Wages policy is inflation policy. The Accord was the first attempt to coordinate rather than legislate. When the dollar fell a third in 1985, policy tightened, and one of the deliberate responses was to discount the depreciation out of wage indexation rather than let imported prices flow automatically into wages. An exchange rate shock answered partly with wages policy. Nobody thought that strange.
Increases have to be funded. The two tier system of 1987 required rises above the safety net to be bought with identified workplace change. There was a counterparty who owed something concrete. The offsets were often notional, and the 1990s productivity surge is usually credited to tariff reduction, competition policy and technology rather than to them. The point is not that the mechanism worked well. It is that it existed, it tied nominal outcomes to real capacity, and nothing replaced it.
And none of it works without a nominal anchor. The inflation target arrived in 1993 and was formalised in the 1996 Statement on the Conduct of Monetary Policy. It came after the decisive disinflation rather than before it, and what it did was turn a victory into a durable regime.
Then we had a team
By the late 1990s almost every major macroeconomic institution had acquired a rule of its own, and most of those rules pointed in the same direction. A target and a mandate for the Bank. Productivity offsets for wage setting. The Charter of Budget Honesty and the convention of surplus over the cycle. Tariff reduction and National Competition Policy for the supply side, with payments to the states to make them do it.
The statutes were the smaller part. The larger part was a set of expectations nobody wrote down. That an increase had to be funded. That a deficit needed a justification. That the target was not negotiable. That competition policy was ongoing work rather than a finished program.
Those understandings were the load bearing part, and because they were never legislated they could go without anyone repealing anything.
Then the teammates drifted off
Competition payments to the states ran from 1997-98 and concluded after 2005-06, with the final assessment under the National Competition Policy arrangements made in 2005. No successor program. Trend productivity growth peaks around 2002 and begins its decline from about 2005.
The surplus over the cycle convention lapsed without a date, because conventions do not have one. The Charter is unamended and governments comply with it. What it actually requires is that a fiscal strategy statement be published. It prescribes no number. A government can comply completely while stating a strategy that constrains nothing, and structural deficits can run across the forward estimates without any rule being broken, because there is no rule.
The productivity offset requirement did not survive the move to enterprise bargaining. An agreement today has to pass the better off overall test, be genuinely agreed, and meet the National Employment Standards. Productivity is not among the things it has to satisfy. The Fair Work Act still lists it among the matters the Commission must have regard to, and having regard to something is not the same as being funded by it.
By 2020 the whole architecture looked unnecessary. Secular stagnation, a Phillips curve that looked flat, neutral rate estimates falling everywhere, and Australia undershooting its own band for most of 2015 to 2019. The Bank acted on that belief. It set a target for the three year bond yield and guided that the cash rate would not rise until inflation was sustainably inside the band, which it did not expect before 2024. The yield target was abandoned in November 2021 and the guidance became the central charge against the Bank in the review that followed.
Once inflation stopped looking like a live constraint, institutions began pursuing their own objectives again. The state wage caps came off. Care sector increases were funded from Canberra rather than earned in the sectors paying them, with no productivity link, by construction.
And in December 2023 the Treasurer and the Governor signed a new Statement. It hardened one thing, making the midpoint explicit where the band used to be the target. It loosened another. Price stability and full employment are stated as dual objectives with no precedence, and on how quickly inflation should return, the appropriate timeframe "depends on economic circumstances and should, where necessary, balance the price stability and full employment objectives."
What was softened is not the target. It is the horizon.
Each of these arrived in its own news cycle, in its own portfolio, with its own minister and its own perfectly good reason. No two of them were ever in the same room. Nobody removed the productivity offset requirement in order to cause inflation, or let the fiscal convention lapse as an inflation policy. Every cut was reasonable on the day it was made.
The machine as it stands
Trend productivity growth, on a ten year moving average, ran near 1.2 per cent in the early 1990s, peaked around 2.4 in 2002, fell to about 1.3 by 2010 and is now 0.29. The pandemic amplifies the recent fall and the non market sector, at minus 0.7 per cent, drags the aggregate. Neither changes the shape. Most of the decline happened between 2002 and 2010.
The architecture worked in an economy growing productivity at over two per cent. The wage outcomes it could absorb then are not the wage outcomes it can absorb now. Nothing about wage setting has to change for wage setting to become inflationary. The room narrowed underneath it.
There is no wage explosion, and anyone claiming one is not reading the release. The Wage Price Index rose 3.2 per cent over the year to the June quarter, down from 3.4, having peaked at 4.3 in December 2023. That is below non tradables inflation. What has changed is composition. Public sector wage growth has outpaced private for six consecutive quarters, and the ABS names state government public service increases as the main driver. The administered part of the wage bill is the part running fastest, in sectors where product market discipline is weak or absent.
Fiscal policy now writes CPI components directly. A cash transfer can deliver the same dollar of assistance without touching the measured electricity price. A credit applied to the bill mechanically lowers the index while it is in place. They chose the second, and Treasury told a Senate committee in November 2024 that the relief had reduced headline inflation by 0.3 percentage points over the year to September. That was offered as a benefit.
And the pressure is domestic. In November 2019 non tradables inflation was 1.9 per cent and tradables 1.8. In July 2026 non tradables is 4.4 and tradables 1.7. Imported inputs sit inside prices we classify as non traded, so the split is not clean. But whatever started this, what remains of it is being generated here.
Nobody does the calibrating
Award wages are set by a tribunal on an annual cycle reasoning partly from the cost of living. State governments set public sector wages. The Commonwealth funds work value increases. Pensions index to the higher of two price measures and are then benchmarked to male average earnings. Fuel excise indexes to the CPI twice a year and tobacco excise to average weekly earnings. Regulators set utility prices. Treasuries decide which prices to subsidise and when to stop.
Every one of those settings is chosen by a different body, for a different purpose, on a different clock. Treasury forecasts the interactions. The Bank models them. Cabinet weighs them. None of that is the same as owning them. No institution carries responsibility for whether the settings cohere, and none is accountable when they do not.
There is a good reason for the rule that says the Bank owns inflation. Accountability norms work by being unconditional, and a central bank permitted to point at Canberra will point at Canberra every quarter. But the rule has a cost nobody counts. If the Bank owns inflation entirely, then wage setting, indexation, administered prices and productivity are not inflation policy. They belong to other departments, answerable for other things. Nobody guards them, because the rule says the inflation architecture is one building in Sydney.
The rule is not a bystander to any of this. It is one of the reasons nobody objected.
The missing function
The obvious institutional answer is another coordinating body. That is also the wrong answer. The whole point of the 1990s settlement was to stop governments leaning on the Bank, and a central bank sitting on a committee with the Treasurer is a central bank that can be leaned on.
It cannot be the Accord either. That required a peak union body with authority over its affiliates and the coverage to make undertakings stick. Union membership is now around an eighth of employees. There is no counterparty to sign anything.
What survives both objections is much smaller. Somebody with the job of publishing an assessment of whether the settings cohere. Whether the wage determination system, the indexation arrangements, the administered price decisions and the fiscal stance are consistent with the inflation target, and saying so when they are not. Directing nothing. Setting no wages, writing no budgets, advising the Bank on nothing.
Independence survives untouched, because nobody is telling the Bank what to do. The assignment rule survives untouched, because the Bank still owns the outcome. What changes is that the decay becomes visible while it is happening rather than twenty five years later in somebody's chart.
That is a modest ask for a problem this size. It is also the only kind of ask that does not require rebuilding something we have already established cannot be rebuilt.
What it costs
Monetary policy will get there in the end. It always does. What the rest of the system determines is how much unemployment it takes to arrive, and monetary policy in Australia works mainly through mortgaged households. The business investment channel is thinner here than the textbook assumes. Manufacturing is a small share of output, mining capital spending answers to global commodity prices and long project cycles rather than to the domestic cash rate, and a great deal of small business credit is secured against residential property.
So the bill for a decayed architecture is not paid by the architecture. It is paid by whoever holds a mortgage and whoever loses a job, and they are selected by when they bought rather than by any principle connected to inflation.
None of this is a forecast. The regression is partial. The target survives, independence survives, nobody has proposed wage indexation. But it is continuing, and nothing in the present arrangement is designed to stop it, because no single cut belongs to anybody's inflation portfolio.
Thirty years of good outcomes taught us the wrong lesson. We were not competent. We were fortunate, and we spent the difference.
Nobody can tell you whether the next twenty years will be as kind as the last twenty. That is the argument for keeping the architecture, not for relaxing it. And we have quietly lowered the level of pressure at which ordinary conditions become a problem.
The Bank is the biggest brick in the wall. It is also the only one anyone inspects.
Inflation was never beaten by one institution getting it right. It was beaten when a team of them, each pursuing its own mandate on its own clock, pulled the same way for long enough to matter. That arrangement is now coming apart.
No comments:
Post a Comment