Build the plant for the trough, not the forecast

Every few years Australia's building-products industry is invited to believe that the housing cycle has been abolished. The invitation usually arrives as a forecast. In 2021 it came from federal home-building grants and record approvals for detached houses. Since mid-2024 it has come from the federal goal of 1.2 million new homes in five years, which implies about 240,000 a year, well above anything the country has built in a sustained run. Investor presentations across our sector now carry a chart of that line rising above the industry's actual output, and the gap between the two is usually described as opportunity.

I understand the appeal. The housing shortage is real, governments of every colour want it closed, and a manufacturer that cannot supply bricks, plasterboard and windows when builders need them loses customers it rarely wins back. But I have spent eighteen years planning capacity at Marrangal, and I have watched more value lost by building for a peak than by missing one. My argument is easy to state and uncomfortable to apply. A building-products plant should be approved on what it earns at the bottom of the housing cycle, and a board should treat any forecast that removes the bottom as a reason for caution rather than confidence.

A goal is not an order book

A housing goal is a statement of what the country needs. It says nothing about what builders will start, what buyers can finance or whether there are enough trades to do the work. Commencements depend on interest rates, construction costs, planning decisions, the solvency of builders and the patience of families waiting for land titles. In the two years after the goal was announced, approvals ran well below the rate it implied, and insolvencies among residential builders stayed stubbornly high. None of that makes the goal wrong. It makes it a poor foundation for an asset that will run for twenty-five years.

Our own history is instructive. Marrangal has owned brickworks through four downturns since the late 1980s. In each, detached commencements in our states fell by between 20 and 35 per cent from peak to trough, and in each the downturn lasted longer than our budget had assumed. A kiln commissioned in a boom typically spends its first years in the downswing that follows. If its business case works only at boom utilisation, it erodes value for most of its working life and then looks heroic for eighteen months at the next peak.

Acquisitions follow the same logic. In late 2021 a well-run roof-tile maker came to market at 11.5 times earnings that reflected grant-fuelled demand we could see would fade. I advised our board to bid no more than A$300 million; another buyer paid A$395 million. The business is still well run. But its earnings, as its new owner reports them, fell by roughly 40 per cent over the following two years, and the volume that was meant to justify the price has not arrived.

What a kiln must earn at the bottom

At Marrangal every capacity proposal now carries a trough case, which the board approves each year as part of its strategy review. It assumes a 25 per cent fall in detached commencements from the latest year, held for two years, with real prices flat and energy priced at the forward contract rather than the spot price we might hope for. A proposal passes if it earns at least its cost of capital across a cycle that includes two trough years, and if, on its own, it would not push us towards our banking covenants in the trough.

The test has changed real decisions. It stopped a second Queensland brick kiln that looked compelling at 2021 volumes. It led us to close the oldest kiln at our Bendarra works in Victoria instead of spending A$31 million to reline it, because at the gas prices available after 2022 that kiln could not cover its cash costs at trough volumes, while two newer kilns nearby could absorb its output. It also shaped the second plasterboard line in Brisbane, which passed only because it replaced board our joint venture had been importing to cover Queensland shortfalls. Displacing imports is volume a company already has; adding capacity in the hope of a policy line is not.

The objection I hear most often is that a trough case makes a company too cautious to meet demand when it comes. The worry is fair, and in 2021 we did run short of some brick colours and window sizes. But the cost of running short in a boom is usually lost margin on the last increment of volume, which builders forgive within a season. The cost of overbuilding is a fixed asset that depresses returns for a decade, which shareholders do not forgive at all.

Capacity that can be switched off

If a company should not build for the peak, how should it serve one? My answer is to separate the capacity a business must own from the capacity it can borrow, restart or buy when the cycle turns.

Owned capacity should be the most efficient in the network and sized close to mid-cycle demand. Borrowed capacity comes from supply agreements with other manufacturers and, for products light and standard enough to travel, from imports. Restartable capacity is the most undervalued of the three. In 2009 we mothballed rather than closed a Queensland concrete-block plant, keeping its permits, its moulds and enough maintenance to bring it back. Holding it cost about A$1.4 million a year. When demand recovered in 2011 it was producing within ten weeks, for a fraction of what new capacity would have cost. I would now pay that kind of holding cost deliberately for one site in each state, and explain it openly to investors rather than hide it in overheads.

Buying in the trough completes the approach. Assets are cheapest when the cycle has squeezed the optimism out of their earnings. We bought Kurrawong Frame & Truss in July 2024, when approvals were weak, for 5.8 times its earnings. The price reflected a business running at about 70 per cent of its capacity. By FY2026 its EBITDA had grown from A$16.5 million to A$19 million without new investment, because the capacity was already there and we only had to fill it.

For directors the questions are practical. Ask management to show each capacity proposal at trough volumes and forward energy prices. Ask which plants the company could mothball rather than close, and what keeping them restartable would cost each year. Ask how much of the next five years' growth in the plan comes from a policy line rather than from customers' order books. And when the investor presentation shows a gap between the federal goal and what the industry actually builds, ask who will be holding the capacity when that gap closes from the other direction.

None of this is an argument against building. The country needs homes, and it needs the materials to build them. It is an argument for plants that are still earning their keep when homes are being built more slowly, which, on Marrangal's experience, is most of the time.