Polyram 2025: Goodwill, EBITDA Multiples, and the Accounting Cushion

Where the Cushion Really Sits

The main article focused on the fact that Polyram widened its global footprint faster than margins recovered. This follow-up isolates the accounting question left underneath that story: how a company that finished a weaker profitability year, and added another acquisition along the way, still ended the year with no goodwill impairment.

This is not a side issue. At year-end 2025, goodwill stood at NIS 284.2 million versus NIS 270.4 million a year earlier. The Lapo acquisition alone added NIS 14.7 million of goodwill and NIS 19.6 million of customer relationships, while also creating a further estimation layer through the Call, Put, and contingent consideration. In other words, the balance sheet became more acquisition-heavy, so the impairment test matters more, not less.

The good news for management is straightforward: every tested unit still shows recoverable value above carrying value. The more interesting question is where that gap actually comes from. In the three legacy operating units, the conclusion rests on EBITDA multiple methodology using comparable companies, with a 10% control premium and 2% selling-cost deduction. In Lapo, by contrast, the test rests on value in use using a DCF with a 15.7% pre-tax discount rate and 3% growth.

The table and chart sharpen the central point. The cushion exists, but it is not evenly distributed. Bondyram alone contributes NIS 271.1 million of headroom, more than the other three units combined. So anyone trying to decide whether Polyram's goodwill test is truly conservative has to start there.

Bondyram: Most of the Comfort, Most of the Sensitivity

In Note 11, the company shows a 12.27 average EBITDA multiple for Bondyram. This is not an abstract figure. It is built from five disclosed observations in the valuation report: Polyram at 9.94, LyondellBasell at 8.76, Mitsui Chemicals at 9.58, Arkema at 4.98, and Asia Polymer at 28.07. The arithmetic average really is 12.27, but the group is clearly not homogeneous. Four of the five observations sit below 10, and one jumps to 28.07.

This is not criticism of the use of market multiples as such. It is a question about the quality of the headroom created by that method. When a unit gets an average multiple that is materially pulled by one extreme observation, the real question is no longer only whether headroom exists, but how much of it comes from the business and how much comes from the peer-set construction.

The implication is sharp, but not sensationalized. If you take Bondyram's EBITDA from the valuation report, NIS 45.6 million, and apply the reported average, you get NIS 603.5 million of net fair value and NIS 271.1 million of headroom over book value. If you remove only Asia Polymer and leave the other assumptions unchanged, the average drops to about 8.32 and net fair value would fall to roughly NIS 409.1 million. Even then, no impairment would have been required, but the cushion would shrink to only NIS 76.7 million.

That is the difference between a test that passes and a test that looks very comfortable. Bondyram still would not drop below book value under a more conservative read, but most of the excess cushion would disappear. So the right sentence on Polyram is not that it has no cushion. The right sentence is that it has a cushion, but a much thinner one than the headline table first suggests.

One more detail matters. In 2024, Bondyram's reported average multiple was 9.81. In 2025 it rose to 12.27. Over the same period, the break-even multiple rose from 5.08 to 6.74. So both book value and the required threshold moved up, and the company still passed the test. But in 2025 more of that comfort clearly depended on a richer multiple environment.

The Other Units Pass, but They Do Not Tell the Same Story

Engineering compounds look more conservative. The average multiple is only 7.69, against a 6.53 break-even multiple. The spread is not huge, but it still leaves NIS 90.4 million of headroom. That is real headroom, just not one built on an especially aggressive double-digit multiple.

Polytron tells a different story. The average multiple already stands at 13.99, while the break-even multiple is 9.04. This test also passes, but the accounting language is clearly more ambitious. Put simply, Polytron is not getting its cushion from 2025 economics alone. It is getting it largely from a relatively rich selected multiple against a small peer group.

This chart matters because it shows that Polyram's accounting cushion is not just a question of absolute shekel headroom. It is also a question of how far the selected multiple sits above the break-even multiple. In engineering compounds, that gap is relatively tight. In Bondyram and Polytron, it is wider, which means the judgment component is larger.

Lapo: A Large Cushion, but in a Different Language

Lapo does not belong in the same bucket. Note 11 tests it using value in use DCF, not an EBITDA multiple. Goodwill allocated to the unit stands at NIS 14.7 million, and recoverable amount is NIS 201.8 million against NIS 65.4 million of net operating assets attributed to the company. On the face of it, that is a large cushion.

But Notes 8 and 22 add an important layer. The same acquisition also created a Call option valued at about EUR 2.786 million, contingent consideration valued at EUR 627 thousand, and acquisition derivatives measured using forecast EBITDA and a 6.5 multiple. That is not an IAS 36 test, so it is not a one-for-one comparison with the goodwill test. Still, it is a useful internal reference point: in the same year, the multiple language protecting Bondyram and Polytron is materially richer than the multiple the company uses to measure Lapo's acquisition derivatives.

That does not mean Lapo's test is weak. If anything, the point is the opposite. Because net operating assets recorded for the unit are only NIS 65.4 million, there is a large buffer before recoverable value becomes a problem. The meaning is different: the new acquisition layer is currently protected more by distance from book value than by an especially generous market-multiple story. That is a different type of cushion from the one Bondyram carries.

The purchase price allocation reinforces that reading. Note 8 shows that the acquisition assigned NIS 25.0 million to identifiable net assets, NIS 14.7 million to goodwill, NIS 10.5 million to the derivative linked to minority buyout rights, and NIS 2.4 million to contingent consideration. So the Lapo balance-sheet layer is not built on goodwill alone. It is built on several estimation layers, which means 2026-2030 performance matters not only for operations, but also for the credibility of the accounting package around the deal.

Conclusion

Polyram does not currently look like a classic case of goodwill write-down being postponed against the numbers. Every tested unit still clears the threshold, and even a more conservative sensitivity on Bondyram does not easily get you to a conclusion that an impairment was already required in 2025.

But this is also not a clean picture of headroom resting only on operating strength. Most of the legacy cushion sits in Bondyram, and most of Bondyram's comfort sits in an average multiple that is heavily influenced by one outlier observation. That is why the more useful investor conclusion is not "there is no accounting issue," but "there is a cushion, and it matters how it was built."

That also defines the 2026 test. If Bondyram delivers EBITDA that supports the 2025 multiple language, the debate fades on its own. If it does not, the headroom could quickly start to look more accounting-driven than economic. And in Lapo, if the forecasts that feed both value in use and the acquisition derivatives are met, the new goodwill will stay away from the center of the discussion. If they slip, the 2025 acquisition layer will move quickly from growth engine to accounting checkpoint.