On 16 June 2026 the CMA cleared Associated British Foods’ acquisition of Hovis. This is a conventional “failing firm” case, but the clearance is not the interesting part.
A failing firm defence has two cumulative requirements: that the firm — here Allied Bakeries (AB) — would likely exit the market absent the merger (Limb 1), and that there is no alternative, less anti-competitive purchaser for it or its assets (Limb 2).
In its Interim Report of 26 March 2026, the CMA provisionally found that, absent the merger, an alternative buyer would have acquired the AB business in Northern Ireland (NI) and continued competing with Hovis. Then the regulator invited to comment on a potential divestiture of AB NI.
Two months later, the Supplementary Interim Report of 21 May reversed that finding, concluding that the most likely outcome was that ABF would liquidate AB NI rather than sell it (as no other credible buyers were likely to exist). The Final Report accepted the failing firm counterfactual and cleared the merger with no remedy.
Limb 1 was, in theory, the easier limb to prove: AB had lost money for years in a declining market. But Limb 2 was trickier: would anyone pay for AB NI, and if so, would it pay more than its liquidation value?
The logic runs as follows. If a third party would pay more for AB NI than its liquidation value, ABF would sell it; Limb 2 is then not met and a divestiture is needed. If nobody would buy AB NI, or would pay less than its liquidation value, ABF would prefer to liquidate; Limb 2 is met, and no divestiture is required.
To answer that question, the regulator stepped out of its competition role and into one usually left to the market — that of an investor judging how much the business is worth and whether the offers made for AB NI were realistic. In this article we challenge some of the assumptions behind that judgement, because mixing how a sale works in real life with how one works in a “regulatory sandbox” can lead to inconsistencies.
Limb 1: Evidence Suggests Big Losses
On Limb 1, the conclusion is hard to fault. AB had been loss-making for years, the plant bread market is in structural decline, and ABF had worked through a long list of restructuring options without returning the business to profit. That AB would have exited in some form was a reasonable finding.
Two cautions worth mentioning. First, much of the exit case rests on ABF’s own internal documents, including a liquidation model prepared once the merger was already being considered, which examined only “merger or closure” and never a sale to a third party. But the evidence suggested profits were elusive. Second, “exit” is not a synonym for “liquidation”.
What “exit” actually means
Exit means liquidation, or sale, or joint venture or other forms to leave the market. The Supplementary Interim Report’s pivot was not that AB would exit; it was that the most likely form of exit was liquidation rather than sale. That is a far stronger claim, and three things sit awkwardly with it.
Consider ABF’s own behaviour. It subsidised AB through losses for well over a decade and retains the capacity to continue. Firms do not fund a loss-maker that long unless they perceive value the accounts do not capture, brand, customer relationships, intra-group volumes. From an accounting perspective, AB is a failing firm, but from a business one, for ABF, keeping the loss-maker somehow was better than closing it down. If so, liquidation is an option, but not necessarily the obvious default.
Then there is the cost of exit itself. Announcing the closure of a heritage brand, with the redundancies that follow, is not a neutral event for a listed conglomerate, a point one third party raised directly, citing potential reputational harm in the financial markets. That cost never enters the net-cash liquidation model, yet it raises ABF’s true reservation value for avoiding a closure. It makes ABF more willing to accept a low sale price to avoid the reputational hit.
Finally, even in a declining sector sales are a real option rather than liquidation. The footnotes in the CMA’s report record distressed bread assets changing hands rather than closing — Rathbones sold to Warburtons; Jacksons and Roberts rescued from administration. “Liquidation is most likely” was a contestable leap, not a self-evident one.
But given the additional evidence on the losses made by AB, let’s assume that no credible buyer would buy AB as a whole, and therefore Limb 1 was met.
Limb 2: Playing to be Investor
Limb 2 is where the regulator’s investor role becomes explicit and mixing theory and reality creates inconsistencies.
The test asks for a less anti-competitive purchaser, not a higher-paying one. Because liquidation here carried a net cost, the CMA reduced it to a question: would a sale leave ABF better off than winding down? The CMA concluded in the preliminary report that for the whole AB there wouldn’t be any buyer, so liquidation was the only option. But for AB NI, since it shows profits, a sale to another buyer was actually possible.
A standalone NI operator buying AB NI is, almost by definition, less anti-competitive than folding it into the merged entity (the parties would hold reportedly around 70% of the market).
When the CMA decided to dig on potential bidders, there were initially 17 candidates, finally down to four, non-binding bids for AB NI. Three of the four came in below the liquidation floor.
The CMA’s report treated that as proof that no buyer would pay above the liquidation value (so the only option was liquidation, no credible alternative). But consider how those bids were produced. The buyers knew the seller had effectively announced it must exit (it was in the preliminary report). There was no rival bidder visible, no second round, no mechanism to counter. A rational buyer in that position anchors low: a non-binding indicative bid against a known seller floor measures the buyer’s read of the seller’s weakness, not the maximum it would pay. Anchors are not ceilings. Imagine that you go to an auction for a house, would you place your first bid as your highest bid? Probably not, but it doesn’t mean you can’t improve your first bid. This is the first inconsistency between the real market scenario and the sandbox one the CMA was playing.
Another problem is where those bids came from. ABF had never run a real sale process and asserted no buyer existed. In other words, because there was never a sale, and nobody showed interest, it was assumed, nobody would buy a “loss-maker” company. But the four appeared only after the CMA published its provisional NI finding and tabled a divestiture remedy. The remedy summoned the bidders whose low offers were then rejected. In other words, it was first accepted that “no buyers” would come using assumptions, and when they came, the offers were rejected using assumptions that nobody would pay more. This is the second inconsistency between the real market and the regulatory sandbox.
Then, there is another argument. The counterfactual is meant to capture what would have happened absent the merger, the world as it stood before the review opened. On that footing the CMA rejected an argument it found inconvenient: that once a divestiture was on the table, the comparison should change (i.e. ABF would be more willing to find a buyer for AB NI just to close the deal). But the CMA held that anything flowing from the merger or its remedy cannot enter the counterfactual.
But the offers it used to rule out an alternative buyer only exist because the review opened. In other words, the CMA or ABF couldn’t find real buyers in the sandbox exercise, but they arrived after the CMA published its provisional finding and floated a divestiture. Investors want real possibilities, not mere exercises. And yet, the offers were non-binding, which also differs from a binding offer.
In other words, the CMA should have either completely ignored the offers from the beginning (as they came after the preliminary review), saying that there were no real bidders (but this was obviously untrue as the process has shown) or let the bidding process continue its natural route, with counteroffers, full due diligence and final, binding offers to see if any bidder would place a bid above the liquidation value. But the CMA can’t have it both ways.
The honest limits
None of this proves a properly run auction would have cleared the floor. Maybe nobody would have offered a price above liquidation. But a process incapable of revealing the true willingness of the bidders to pay cannot establish that no buyer existed above liquidation value.
In any event, even if Limb 2 failed there, the merger would not fall: the consequence would be a divestiture, the rest of the deal proceeding on the exiting firm basis. The Hovis transaction would survive. What would change is whether one regional bread market kept an independent competitor that the market, had it truly been allowed to speak, might have preserved on its own.
The lesson is not that the CMA’s financial analysis was weak — it was strong. It is that strong financial analysis needs to take into account the market’s realities, and not just focus on the academic exercise. A reservation value is an input to a negotiation, not a verdict on it. When a regulator must infer what the market would do, the safest evidence is a real process.
