Decisions that changed what Corvalla owns and how it builds

Selling four office towers before the market recovered

When I became chief executive in July 2021, office towers in Melbourne, Sydney and Brisbane made up 54% of Corvalla's A$8.1 billion balance sheet. Occupancy in our Melbourne towers had fallen to 86%. Interest rates then rose by more than four percentage points between May 2022 and November 2023. By December 2022 our gearing had reached 36.1%, above the top of the board's 25–35% range, and our securities traded about 30% below net tangible assets.

The question was whether to sell towers into a weak market or hold them until values recovered. Our chief financial officer, Andrea Kostelic, argued for holding two years and paying down debt from retained earnings, and two directors agreed with her. Our largest securityholder, Orsolin Asset Management, with 9% of the register, wanted any proceeds spent buying back securities at the discount. I proposed selling four of our eleven towers over three years, using the money first to bring gearing back inside the range, then to fund logistics and housing, with a buyback only once gearing fell below 32%.

The board approved the programme in February 2023. We sold the four towers between June 2023 and May 2025 for A$1.46 billion, on average 8.7% below their June 2022 book values and within 3% of the reduced values after our December 2022 write-down. We gave up income: funds from operations per security fell from 25.1 cents in FY22 to 21.4 cents in FY24, and we reset the distribution from 20.4 to 17.1 cents.

By June 2026 gearing was 29.8% and office was 32% of the portfolio. The proceeds funded 186,000 square metres of logistics development at a 6.4% yield on cost and our share of two rental buildings, and we bought back A$90 million of securities in FY25. Funds from operations per security recovered to 22.6 cents in FY26.

The annual hold-sell review that began with this programme now covers every asset. I would change two things. I announced the distribution reset six months after the sales programme, so investors absorbed bad news twice. And I sold our Brisbane tower first because it was the easiest to sell; Brisbane recovered sooner than our other markets, and a comparable building traded eighteen months later for about A$28 million more than we accepted.

Commissioning valuations our own policy did not require

In 2022 Corvalla's policy was to value each property independently at least once a year, with directors' valuations in the half-years between. At December 2022, six of our eleven office towers were due for directors' valuations. CBD office transactions had nearly stopped, so little evidence contradicted the values we held, and our internal work suggested a fall of about 4%. I did not believe it. Incentives on new leases in our Melbourne towers had reached 38%, and renewing tenants were taking on average 22% less space than before.

I asked the board to commission independent valuations of all eleven towers for the half-year accounts. Andrea Kostelic, our chief financial officer, opposed it: the policy did not require it, valuers would have little transaction evidence, and a larger write-down could unsettle lenders. Margot Faircloth, who chairs our audit and risk committee, asked first for an analysis of our debt covenants, which showed gearing would rise but stay well inside the 50% covenant limit. The board agreed to the off-cycle valuations in November 2022.

The valuers reduced the office portfolio by A$412 million, 9.6% of its book value. Net tangible assets fell by 29 cents per security, and our security price fell 6.8% on the day of the half-year result. In the same result we began publishing, for every tower, face rent, incentives on new leases and net effective rent. Several analysts told us they had been estimating those numbers for us and getting them wrong.

During 2023 our lenders extended A$1.1 billion of facilities without changing covenants, and two said the valuation decision had helped. When we sold four towers between 2023 and 2025, prices settled within 3% of the written-down values, which suggested the valuers had been close. Our policy now requires independent valuations every six months for any office tower with vacancy above 10% or incentives above 30%.

What I would change is when we spoke. At our annual meeting in November 2022, before the valuers had reported, I described office conditions in general terms. I should have told securityholders then that we had commissioned off-cycle valuations of every tower and why, so the result did not arrive as a surprise.

Replacing a fixed-price contract on a Parramatta rental tower

In mid-2022 Corvalla and Skarholt Pension Fund, our equal partner in the build-to-rent platform, tendered the construction of a 410-home rental tower in Parramatta. Our cost plan put construction at A$298 million. We invited six builders to bid a fixed-price, design-and-construct contract, and four declined. The two that bid priced the work at A$349 million and A$352 million, both excluding steel and concrete price movements, and one told me privately that it would not sign a fixed price for a tower of that size again. Several builders had failed across the country that year, and I took the warning seriously.

The choice was between a fixed price 18% above our estimate that still excluded the risks we most feared, or a different contract. I proposed a two-stage contract with Ferngate Construction: a paid first stage to finish design and tender the main trade packages on an open book, then a stage-two price with a fixed builder's margin of 4.5%, savings and overruns shared equally within a 5% band, and the owners carrying latent ground conditions and price movements on four major packages. Andrea Kostelic and Skarholt's representative on the project committee both preferred the certainty of the higher fixed bid. Skarholt agreed only after we capped our joint exposure to price movements at A$14 million and committed to monthly cost reports.

We gave up a fixed number. Development contingency rose from 5% to 8%, and the board received a project risk report every quarter. The stage-two price, agreed in March 2023, was A$321 million. The final cost was A$327 million, including A$9 million of price movements that we carried. The building opened in May 2025, three months behind the original programme, and was 94% leased by February 2026.

The two-stage model now governs every Corvalla housing tower; repeatable warehouses still use fixed prices. When we tendered Footscray in 2024, five builders bid. What I would change is the band. Most of the overrun came from the facade package, which we left inside the shared band although its design was the least resolved at the end of stage one. It needed its own price and contingency, and our cost-management team should have been strengthened before stage one, not halfway through it.

Redesigning a refused Footscray scheme instead of appealing

In early 2021, while I was chief investment officer, Corvalla bought a 6,400-square-metre site near Footscray station for A$41 million to build rental homes. Our first application, lodged in 2022, proposed 19 storeys and 402 homes. In April 2023 the council refused it, citing height, overshadowing of a public park and the abrupt transition to the two-storey houses in the street behind. Our development director, Callum Delahunt, recommended an appeal to the Victorian Civil and Administrative Tribunal. Our advisers put the chance of success at about 60%, with a decision in fourteen to twenty months.

I had written refusals like that one as a council planner. Read carefully, its specific grounds described a building that could be approved. I proposed a redesign at 13 storeys, stepping down towards the houses, with the tallest part moved away from the park. Callum argued that we would be giving up homes the city needed, and value we had paid for, on the strength of my reading of one document. Skarholt, our partner, was more worried about delay than about height. The board accepted the redesign in June 2023.

We gave up 118 homes, taking the project to 284, and wrote the land down by A$7 million. The forecast unlevered return fell from 7.4% to 6.6%. Before lodging again we met council officers three times and held two sessions with neighbours, and we agreed with Wattlecourt Housing, a community housing provider, that 43 homes would be let to key workers at a 25% discount to market rent for fifteen years.

The council approved the amended application in December 2023, five months after lodging. Construction began in September 2024, and the building is due for completion in December 2026. Had an appeal succeeded at the midpoint of our advisers' estimate, construction would have started roughly a year later, into a market where building costs were still rising.

No significant Corvalla application is now lodged without pre-lodgement meetings, and every planning strategy states in advance when we would redesign rather than appeal. What I would do differently is plain. As a former council planner, and someone who had made pre-lodgement meetings routine at Yarrendale, I should have insisted on them before the first application. We had designed for the rental market and the site's price, and the street behind had to tell us what we missed.