In 2025 Renault reported a net loss of €10.9bn, of which €9.3bn arose from a change in the accounting for its investment in Nissan. However, no shares were sold and nothing happened in 2025 to the value of the stake to justify a loss of this size.
The charge results from reclassifying Nissan from an associate, measured using the equity method, to a financial asset measured at fair value. A similar accounting construct, this time a gain, appears in Microsoft’s 2026 results, arising from dilution of its stake in OpenAI. Both reveal, in our view, the deficiencies of equity accounting.
Renault’s initial purchase of 36.8% of the shares of Nissan occurred in 1999, but its influence over Nissan has been progressively reduced in recent years. Under the New Alliance Agreement of November 2023, Renault transferred most of its Nissan shares into a French trust, with voting rights capped at 15%, and subsequently sold part of its holding, taking its interest from 43.4% to 35.7%. Further changes in 2025 to the terms and conditions governing Renault’s rights led the company to conclude that, from 30 June 2025, it no longer exercised significant influence over Nissan.
The consequence under IFRS is that, without significant influence, IAS 28 equity accounting no longer applies. Instead, the stake becomes a financial asset within the scope of IFRS 9, and is therefore measured at fair value. The accounting effect of crossing this threshold was dramatic: the investment was written down from its equity-accounted carrying value to fair value based on Nissan’s quoted share price, with the difference recognised in profit and loss. The resulting charge of €9,315m was included in other operating income and expenses and is part of the €9,407m loss shown in the note below (the remainder relates to a small sale of Nissan shares during the year).
Extract from Renault ‘Other operating income and expense’ note
Renault 2025 financial statements
The explanation provided by Renault reveals the complexity of this reclassification adjustment. The dominant component is the difference between cost and fair value, but there is also the recognition of the cumulative translation differences, which were previously reported in OCI in accordance with the usual ‘closing rate method’ of currency translation, and a reversal of a related deferred tax balance. We do not think that any of these three components of the overall loss has any economic significance in assessing performance in 2025. They all arise from the change in measurement basis and not from any value or performance relevant event occurring in the period.
Crossing an accounting threshold does not change economic value
Renault is not alone in reporting large gains or losses when investments cross accounting boundaries. Whenever a stake moves between subsidiary, associate, and ‘plain’ equity investment classifications, the change in measurement basis can itself produce a gain or loss, even if the holding and its value are unchanged, or if the transaction is value neutral. We described a similar effect, also involving a loss of significant influence, in our earlier article about the Vivendi stake in Telecom Italia1See our article ‘Impairment of associates may not reflect underlying economics’..
An accounting loss, but not an economic loss
The €9,315m charge is perhaps best understood as the recognition of a difference that had accumulated over many years: the gap between the equity-accounted carrying value of Nissan and the fair value of the stake, plus historical currency translation differences that had previously bypassed profit and loss.
Equity accounting is essentially a form of cost-based measurement. The carrying value of an associate is the cost of the investment plus accumulated retained earnings (both net income and other comprehensive income). We regard the increase in carrying value from accumulated retained earnings as also being a form of additional investment at cost, in the sense that the investor has not received a full distribution of earnings.2We explain more about the historical cost nature of the equity accounting measurement of associates in our article ‘Equity investments – Why accounting differences matter’.
The equity accounting ‘cost’ can move a long way from fair value. In Renault’s case the change was extreme. At the end of 2024, Renault disclosed that the stock market value of its Nissan stake was €3,904m, some 69% below the €12,599m carrying value in the balance sheet. A significant gap had already existed at the end of 2023.
Extract from Renault’s investment in Nissan note
Renault 2025 financial statements
The full effect of Nissan on the Renault 2025 income statement is a charge of €11,646m. This combines the €9,315m reclassification loss with the €2,331m share of Nissan losses recognised under equity accounting prior to the change on 30 June 2025.
The decline in the value of Nissan happened over several years and was clearly visible to investors well before 2025; Nissan is listed and its share price is observable daily. The reclassification loss reported in 2025 does not provide new information about performance in that period. It is the belated recognition in profit and loss of value changes that occurred, and were already known to the market, in earlier periods.
Renault itself was clear on this point. The company’s announcement emphasised that the charge is non-cash, does not affect the dividend, and does not change the strategic and operational cooperation between the two companies. We agree; shareholders did not become €9.3bn worse off in 2025 because of an accounting event that happened in that year.
The 2025 loss primarily reflects value changes from earlier periods
Financial statements inevitably include some catch-up adjustments, but a catch-up of this size raises an obvious question: if the loss tells investors nothing about current performance, what is the purpose of reporting it in profit and loss at all? In our view it provides no useful information, and its prominence in the 2025 income statement risks obscuring the underlying result. Renault was careful to present net income excluding the Nissan effect, and we suspect most analysts focused on that.
Where was the impairment?
If the fair value of Nissan had fallen so far below carrying value, why was the difference not recognised earlier through impairment? An impairment of €694m was recognised in 2024, but this barely dented a gap of almost €9bn.
The answer lies in the mechanics of the IAS 28 and IAS 36 impairment test. An associate is written down only when its ‘recoverable amount’ (the higher of value-in-use and fair value) is below the balance sheet carrying value. The problem is that value-in-use is based on management’s own cash flow forecasts and discount rate assumptions, and in practice gives considerable discretion to avoid or minimise write-downs. Renault’s 2024 test, using an 8.89% discount rate and management’s medium-term forecasts for Nissan, supported a carrying value roughly three times the observable market value of the shares.
Value-in use-gives management discretion to defer losses
We think this illustrates a wider failure of impairment testing for equity-accounted investments. For listed associates in particular, we see no good reason why the observable market value of the shares should be overridden by a management-derived value-in-use. A test anchored to market value would have recognised the decline in Nissan progressively, in the periods in which it actually occurred, rather than in a single artificial charge triggered by a classification change. Renault is not an isolated case; there are several current situations in which the carrying value of a listed associate substantially exceeds the market value of the underlying shares, such as the investment by HSBC in China’s Bank of Communications.
Some argue that the additional value-in-use test has merit because market prices can be temporarily depressed and a long-term holder with broader insight may legitimately have a different view of fundamental value. However, in our view, it lacks the discipline of fair value, especially where market prices are available.
Another problem – dilution
It is not just the reclassification of an associate to an investment measured at fair value that can result in entries in profit and loss that are difficult to explain and fail to reflect underlying economics; so too can the issue or repurchase of shares to or from a third party. Even though the investor’s absolute holding does not change, and the actions of the associate are essentially value neutral (the shares are issued or repurchased at fair value), the investor can still report a gain or loss.
The mechanics are best seen in a simple example. Suppose a company has a 20% stake in an associate that has shareholders’ equity of 100. Ignoring complications such as goodwill, the effects of a purchase price allocation,3In our example, we assume no goodwill is included in the carrying value of the equity accounted investment. This would commonly happen where the investment was obtained at the incorporation of the business. If acquired later, it is likely the carrying value will include goodwill in addition to the share of shareholders’ equity. A dilution gain arises when equity is issued to another investor at a price higher than the book value per share reported under equity accounting, inclusive of the effects of goodwill and other adjustments that arise from the equity accounting purchase price allocation. and unrealised profit adjustments, the investment in the associate would be recognised at 20. Assume the market value of the associate is 1,000 and the associate decides to issue additional equity. If the number of shares is increased by 10%, with the new shares sold at fair value, this raises 100 and increases shareholders’ equity to 200.
Value neutral in accounting, but a gain in economics
If the investor does not participate in the issue, their holding falls to 18.2% (20/110). However, under equity accounting, the carrying value of this investment is now 36.4 (18.2% of 200). The increase in carrying value of 16.4 is reported as a gain in profit and loss. In value terms nothing has changed; 20% of 1,000 is the same as 18.2% of 1,100. However, in accounting terms a profit is recognised (or a loss is recognised if the market value of the associate were less than balance sheet shareholders’ equity).
The ‘profit’ arises because new equity is sold for a higher per share value than the per share book value reported under equity accounting. This can be thought of as a type of ‘opportunity gain’, in that new equity is sold at a higher price than that paid by the equity accounting investor, therefore increasing their per share book value – the original investors are better off having delayed the issue of further equity. It can also be viewed as simply a partial catch-up of book value to be closer to fair value.
In whatever way it is rationalised, we think a dilution gain or loss is difficult to relate to the underlying economics and is irrelevant when assessing current period performance.
To illustrate that this is a real issue in practice, we show below an extract from the Microsoft annual report for the year ending June 2026.
Extract from Microsoft ‘Other income and expense’ note
Microsoft 2026 Form 10-K, note 3
A reported ‘gain’ at odds with the underlying economics
The “Other, Net” gain of $4.7bn in 2026 includes $6.5bn of gains related to the Microsoft stake in OpenAI, which it accounts for as an associate. The note includes the share of associate net losses, but in 2026 this was offset by a gain arising from an equity issue by OpenAI. Microsoft says that the $6.5bn ‘primarily relates to the dilution gain from the OpenAI Recapitalization’, which included the issue of new equity at a price higher than the book value reported in the Microsoft financial statements, similar to our example above.
In our view, this gain is also an accounting construct that is at odds with the actual economic position of Microsoft’s stake in OpenAI. In fair value terms, the value of the Microsoft investment increased by over $100bn during the same period. The dilution gain may happen to be directionally correct, but we think it is still misleading.
Although Microsoft reports under US GAAP, IFRS produces the same effect, with the dilution effect often referred to as a ‘deemed disposal’ under IAS 28.
Fair value would have told investors more
There are some circumstances where we think a direct focus on operating flows through equity accounting or proportional consolidation, rather than value changes, may be the preferred accounting. This particularly applies to some joint ventures and associates that are integrated with the investor’s operating activities. Nevertheless, both the Renault and Microsoft cases reinforce the view we have expressed before: fair value is generally a more relevant measurement basis than equity accounting for equity investments where the investor does not have control, whether or not there is significant influence.
Had the Nissan stake been reported at fair value throughout, changes in value would have been recognised as they occurred, the balance sheet would have reflected economic reality, and there would have been no artificial €9.3bn charge in 2025 for investors to interpret, and for the company to explain away. The same applies to Microsoft, where the fair value gain in 2026 would have been much higher than the difficult to explain and, we think, misleading, dilution gain.
For Renault the investment is listed, and for Microsoft the extensive funding transactions in OpenAI shares provide observable pricing, both of which enables us to easily contrast the equity accounting gains and losses with fair value changes. However, we also believe that fair value is just as relevant for investments without observable prices, despite the measurement uncertainty inherent in level 2 and level 3 fair value estimates. Investors are still better off with fair values and supporting disclosures.
We prefer fair value through profit and loss, not FVOCI
Although fair value measurement now applies to the Nissan stake, we think Renault’s implementation choice is unfortunate. The company has elected the IFRS 9 ‘fair value through other comprehensive income’ option, under which subsequent value changes are reported in OCI and never recycled to profit and loss. In our view this split of equity investment gains and losses between profit and loss (the dividends received) and OCI (the rest of the total return) is artificial and unhelpful.
We would prefer all equity investments to be measured at fair value through profit and loss, with the gains and losses separately presented in the investing category of the income statement under IFRS 18. This presentation keeps investment returns clearly separate from operating performance while ensuring they are still reported as part of net income.
Insights for investors
- The €9.3bn Renault charge is a consequence of a change in accounting method, not an economic loss incurred in 2025.
- Equity accounting is a cost-based measure; the carrying value of associates can differ materially from fair value. For listed associates we suggest you always check the market value of the underlying shares.
- Impairment testing based on value-in-use gives management additional discretion that can significantly delay loss recognition.
- Watch out for reclassification gains and losses whenever investments cross control or significant influence thresholds; these likely reflect accounting mechanics rather than current period performance.
- Gains and losses also arise when an associate issues or repurchases its own shares; these dilution effects say nothing about changes in the value of the investment.