The half in numbers
- Global direct investment +28% year on year in Q2 (JLL)
- Europe H1 €116bn, +10%; United States H1 $250.3bn, +21% (CBRE)
- European prime office yields +3bp in Q2 — the first rise in two years (Cushman & Wakefield)
- Five-year GBP swap 4.39%, roughly 80bp wider than in February
- US CMBS delinquency 7.86%; office 11.91% (Trepp, July)
- Closed-end real estate fundraising $92.6bn, −38% — the weakest first half in nine years (PERE)
What happened
Capital came back everywhere. Europe and the United States both grew double digits on the half, and Asia Pacific recorded its strongest half on record. Office leasing reached a post-pandemic high.
Cheap debt did not come back with it. The Fed held at 3.50–3.75% in July with three dissents in favour of a rise; the Bank of England held at 3.75%; the ten-year gilt sits above 5%. Values have not followed volumes — European capital values edged lower in Q2, and UK values fell on outward yield movement in logistics, regional offices and residential.
Where the divergence is
Retail is the recovery nobody forecast. UK retail returned 7.8% over twelve months, the best of any sector, and prime yields compressed 25–50bp — the only UK sector to compress. Logistics went the other way, prime distribution yields moving out 25bp even as take-up rose. Office is two markets in one line item — US prime vacancy 12.3% against 18.3% overall, values still 35% below their 2022 peak. Living carries the most capital and the weakest returns: Europe’s largest sector by volume, the worst-performing UK sector, and US multifamily distress more than doubled from 6.0% in February to 13.0% in July. Data centres barely behave like property — 25GW absorbed in the half, double a year earlier, vacancy at 1%.
What the authorities did
No tax authority announced a real estate transfer pricing campaign this half. Several did something more consequential.
In January the Paris Administrative Court of Appeal decided Trema Holding — a €32m Luxembourg shareholder loan on a Paris-region office at 4%, originated in 2006 and extended in 2014. The court accepted the taxpayer’s B1/B+ rating and rejected the authority’s implicit support argument, holding that the lender must be treated as having reassessed the risk at the extension date. It disallowed the rate anyway: nine of ten bonds in the first study were USD-denominated with no eurozone comparability shown, and 2013 bonds were not shown comparable to November 2014 conditions. Deduction was cut to the statutory rate of 2.3%. The credit analysis survived. The search did not — and the date that mattered was the extension, not the origination.
Germany now tests whether the borrower can service the debt across its full term, capping the rate at the group rating. The Netherlands is studying default non-deductibility of intra-group interest funding real estate. In the UK, advance certainty on leverage has quietly gone: HMRC agreed two Advance Thin Capitalisation Agreements in 2024-25, against 45 in 2019-20 — a fall HMRC attributes to the Corporate Interest Restriction. Enquiries now settle at an average age of 41 months.
The ajiho view
Two things are converging on the intercompany layer.
The first is the maturity calendar. $875bn of US commercial mortgages mature this year and $1.26tn in 2027; 19% of the UK book matures this year; AEW puts Europe’s debt funding gap at €74bn across 2026–28, 41% of it offices. Loans struck at 4.1–4.7% are refinancing at around 6.5%. The legacy coupon is not the exposure — a fixed-rate loan priced correctly in 2021 was tested against 2021 and stays arm’s length for its term. The exposure is that a renewal is a new transaction, which is precisely what Trema turned on. A group refinancing external debt this year is almost always resetting the intercompany layer above it — priced against the market on the day it is made.
The second is who the lender now is. Debt funds hold 28% of the UK outstanding book, up from 12% a year earlier; alternative lenders took 38% of US non-agency closings in Q2. Where the borrower’s realistic alternative is a debt fund at 55–60% LTV rather than a bank, the comparable set has moved and the price with it.
The sector spread is what most files still ignore. One group margin across retail, logistics, offices and multifamily asserts that a lender would price four markets the same in a half when they moved in four directions.
Our call. Dispersion widens through H2 2026. Expect the gap between prime and secondary within each sector to exceed the gap between sectors, and expect a single group-wide margin to be the first thing questioned on a real estate file.
What this means for your business
- Treat every 2026 and 2027 refinancing as a repricing event for the intercompany layer, not only the external one. The renewal is the transaction that gets tested.
- Price sub-sectors separately across a diversified portfolio. Retail compressing while logistics widened in the same half is not a rounding difference.
- Establish whether the borrower’s realistic alternative is still a bank. Where it is a debt fund, a bank-priced comparable set describes a market the borrower has left.
- Test the search on currency, date and jurisdiction before you defend the rating. Trema lost on the search after winning on the credit analysis.
Sources
CBRE (European and US capital markets, Q2 2026); JLL (Global Real Estate Perspective, August 2026; data centres, H1 2026); Cushman & Wakefield DNA of Real Estate (Q2 2026); MSCI/IPF UK Quarterly; Green Street Commercial Property Price Index; Knight Frank Prime Yield Guide (May 2026); Trepp via MBA NewsLink (July 2026); CRED iQ (August 2026); Bayes Business School UK CRE Lending Report (YE2025); AEW; Mortgage Bankers Association; PERE. Federal Reserve, ECB and Bank of England policy statements, July 2026. HMRC, Transfer Pricing and Diverted Profits Tax statistics 2024 to 2025. Cour administrative d’appel de Paris, No. 24PA02156, January 2026. Market data as at 21 August 2026.