In February we said Japan's election would test Australian balance sheets within six to twelve months. We were right about the mechanism. We were wrong about the pace and about who else would blink.
On 9 February, the day after Sanae Takaichi's Liberal Democratic Party won its supermajority, I wrote that markets had read the result not as political instability but as permission, permission for higher bond issuance, and a higher long-run cost of capital.
I said the pressure would concentrate in long-dated Japanese government bonds. I also said Japan is not just another bond market but the marginal price setter for global duration. And I said Australia would import tighter financial conditions whether or not the Reserve Bank did anything.
I gave it six to twelve months. We are now six months in.
The thesis of my article has held. The timing has not. It has arrived faster and harder than I expected, and it has picked up a second engine I did not forecast.
What has happened since February
Japan. The 10-year JGB has pushed to around 2.95%, a level not seen since 1996. The 30-year sits above 4%. The government still has not explained how it will fund the two-year cut to the consumption tax on food, the very spending mandate the election delivered. Markets are now openly pricing the possibility of a Bank of Japan hike as early as September.
Takaichi's fiscal programme assumes growth outruns borrowing costs. With the 10-year approaching 3%, inflation near 2%, and real growth forecast at roughly 1%, that assumption does not survive contact with the yield curve. A government elected to spend is discovering the price of the mandate.
The capital flows. The repatriation I described is now measurable. Japanese investors sold roughly US$29.6 billion of US debt in the first quarter of 2026 alone. Foreign holdings of Treasuries declined again in June, with Japan, China and the United Kingdom all reducing. The buyer of last resort for global duration has gone home.
The United States. This is the part I did not forecast. On 19 August, with the 30-year Treasury at 5.26%, its highest in nineteen years. The US Treasury announced it would at least double the size of its debt buyback operations, from US$2 billion to at least US$4 billion, targeting the 10-to-30-year sector that had seen a buyers' strike since late June. Yields fell roughly nine basis points on the announcement. The same week, US public debt passed US$40 trillion for the first time.
The world's deepest, most liquid bond market required official support to clear at the long end. Whatever else that is, it is the clearest available evidence that the pricing power I described in February has really moved.
Australia. Our 10-year is back above 5%. The 30-year is at 5.58%. The three-year sits around 4.49%, above the cash rate, which tells you the market has not finished pricing risk into the front of the curve either.
Where we were wrong
I said Australia would face rising funding costs without an RBA cash rate hike. I got both.
The cash rate is 4.35% after three increases this year. The Bank has held for two consecutive meetings, but Governor Bullock has been explicit that rates may need to rise again. Westpac is forecasting two further increases. The CAMA Shadow Board puts the probability of higher rates over the next six months at around 70%.
I under-called it. Domestic inflation, energy costs and a labour market that will not loosen have done work I attributed almost entirely to imported duration risk. The two forces have compounded rather than substituted for one another. That is a materially worse outcome for a leveraged balance sheet than either alone.
I was also wrong on the currency leg. I said Japan would sell US dollar securities and the greenback would fall. Instead, the yen stayed weak enough that the US Treasury intervened to support it earlier this month, in part to stop Japan needing to liquidate Treasuries to raise dollars.
My February scorecard
| What we said in February | Where it stands |
|---|---|
| Volatility in long-dated interest rates | Confirmed |
| Pressure on rate-sensitive assets — property, infrastructure, private credit | Confirmed |
| Rising funding costs without an RBA cash-rate hike | WrongCosts rose and the RBA hiked. |
| Selective, not system-wide, distress | HoldingConcentrated in construction, development and geared mid-market borrowers. |
Why this matters more than the next RBA decision
Most boards are still watching the cash rate. It is the wrong number.
Three- to five-year fixed business debt, development finance and commercial property lending are priced off the swap curve, not the overnight rate. A long end anchored above 5% raises the cost of refinancing even in a world where the RBA never moves again and it does so on a schedule set in Tokyo and Washington not Australia.
That distinction has four practical consequences.
- Interest cover compresses before revenue does. A facility written in 2022 at a 3-handle and refinancing into a 6-handle is a covenant problem long before it is a trading problem.
- Cap rates follow the long bond, not the cash rate. National property prices fell 0.7% in July, the largest monthly decline since December 2022. Valuations move ahead of the profit and loss.
- Private credit reprices quietly. The stress does not announce itself through a bank. It appears as a lender declining to extend or extending on terms that consume the equity.
- The FY27 refinancing wall is now an FY26 conversation. Anything maturing inside eighteen months should be modelled at today's curve, not at the rate on the existing facility.
What boards should do now
- Reprice the debt stack at the current swap curve, not the current rate. If you have not modelled a refinance at 6% plus margin, you do not know your position.
- Bring the refinancing conversation forward. Lenders are materially more constructive twelve months out than three months out. This is the highest-return action available and it costs nothing but calendar time.
- Test covenant headroom on interest cover, not EBITDA growth. The break usually comes through the denominator.
- Revalue geared assets on a duration basis. If your asset values assume a 4% long bond, they are stale.
- Document the analysis. Where the numbers are tightening, safe harbour protection depends on the plan existing before the pressure peaks, not after.
Final thought
In February I wrote that this would not be a “recession we had to have.” It would be a repricing we need to navigate, and the danger would sit in asset values, spreads and credit decisions rather than in mass unemployment.
I still stand by that. What I would add is that when the world's largest bond markets require official intervention to clear, the repricing is no longer a forecast. It is the operating environment.
