Convertible bonds: US GAAP and hidden financing costs

Most convertible bond issues in the United States are no longer structured to physically deliver shares when converted. Issuers prefer a structure where only the conversion premium is settled in shares. The accounting, rather than the economics, appears to show why.

We explain how the net share settled structure works, how under US GAAP it produces a combination of an unrealistically low interest expense with a mostly zero diluted EPS effect, and why IFRS reporters have not followed. It strengthens the objection we raised to the 2020 changes to convertible accounting under US GAAP.


The issue of convertible bonds by companies in the United States is on course for a record year in 2026, with zero-coupon convertibles comprising about 41% of the total, according to the Financial Times1See ‘Zero-interest convertible bonds set for record year’, Financial Times 28 August 2026. Issuance data attributed in that article to Dealogic and Barclays Research.. The attraction of convertibles as a source of finance is often attributed to the low coupon combined with the conversion premium, where the conversion price (the effective price at which shares are issued if conversion takes place) is significantly above the share price at the time of the convertible being issued.

Convertible bond terminology: Click here if you are unfamiliar with some of the terms we use.

Advocates of convertible financing may characterise it as cheap debt (the low coupon), coupled with cheap equity (fewer shares and less dilution). However, this ignores the optionality inherent in this form of financing (the choice available to investors to convert or not) and the resulting cost to existing shareholders of writing this embedded call option. In economic terms, the cost of convertibles should simply reflect the risk-return characteristics of the security, allowing for both its debt and equity call option components.

Settlement features now common in the US market exacerbate the accounting problems

While understanding the economic cost of convertibles is a challenge for investors, we think the bigger problem is that this cost is misrepresented in financial statements, particularly under US GAAP.

We have written about much of this previously (there is no need to read these before continuing):

The reason we return to the topic now is a development in the US convertible bond market over the last few years whereby ‘traditional’ convertibles, which feature full physical settlement in shares if converted, have largely been replaced by a structure involving partial settlement in cash. An issuer adopting this structure obtains a materially better (from the company’s perspective) financial reporting outcome, in both the income statement and the diluted EPS calculation. In fact, we think the popularity of this structure in the United States is primarily driven by the accounting advantage. IFRS accounting for convertibles is different and better reflects the economics, although it is not without its own challenges, as we explain later.

First, we explain the net share settled structure and illustrate it using a May 2026 issue by US semiconductor company onsemi (yes, it is lower case). We then examine how the new structure impacts the four things we think can cause investors to misunderstand and underestimate the cost of convertible financing:

  • The absence of any bifurcation of the instrument into debt and derivative components under US GAAP and the resulting understatement of the interest expense.
  • The treatment of the derivative overlay – the ‘dilution hedging’ transactions that now accompany almost every issue – whereby a cash payment that is effectively interest expense, is charged directly to equity.
  • The diluted EPS calculation, which under this alternative structure will, at least initially, record no dilution at all.
  • The cost of capital, which is widely misunderstood, irrespective of how the convertible is structured and reported.

The main incremental accounting benefit of the new structure is in the diluted EPS calculation; the first two accounting effects, and the cost of capital challenges, are the same as we have previously explained for traditional physically settled convertibles. We include all 4 in this article for completeness; if you want to focus on what is new, read the next section and then skip forward to read about the effect on diluted EPS. 

Net share settlement and the onsemi convertible

A net share settlement convertible features the same conversion ratio as a traditional convertible that is fully settled by issuing shares, and delivers the same value when conversion takes place. However, settlement is not wholly in shares. That part of the settlement value equal to the par value of the bond is repaid in cash, and the conversion value in excess of this is typically either settled by issuing shares to the appropriate amount, or through a further payment of cash at the option of the issuer.

Fewer shares are issued under this structure, but the issuer must also make a cash payment. From the investor’s perspective the instrument is still a hybrid which combines a bond floor and equity upside, and which has the same expected return; it is just that what is actually received on conversion differs. Of course, for all convertibles, if conversion does not take place investors simply receive repayment at par in addition to the specified coupon (if any).

onsemi zero-coupon convertible and derivative overlay

In May 2026 onsemi issued $1.5bn of zero-coupon convertible senior notes due 2031 on net share settlement terms. The initial conversion rate of 6.1997 shares per $1,000 gives a conversion price of approximately $161.30, which is a premium of 52.5% over the share price of $105.77 on the pricing date. Consistent with our description above, the company will satisfy any conversion by paying cash up to the principal amount converted and delivering cash, shares or a combination in respect of the excess, at its election.

If, for example, investors elect to convert at a time when the share price is $200, the total value delivered will be $1.86bn (1.5bn x 6.1997/1000 x $200). Of this, $1.5bn is delivered in cash and the rest either in cash or shares. If share settlement is elected, 1.8m additional shares ($360m / $200) would be issued. This compares with the 9.3m shares (6.1997 x 1,500m / 1,000) that would be issued under a traditional share settled convertible.

At the same time as the convertible was issued, onsemi purchased convertible note hedges in the form of call options with the same terms as the option embedded in the notes, for $351.6m, and sold warrants over the same number of shares at a strike price of $211.54 (a 100% premium to the then share price) for $281.0m. The net cost of $70.6m was charged to shareholders’ equity because both instruments qualify as equity derivatives.2onsemi Form 8-K dated 12 May 2026 and Form 10-Q for the quarter ended 3 July 2026.

onsemi May 2026 convertible note financing

The Footnotes Analyst, from onsemi Form 10-Q for the quarter ended 3 July 2026

Here is the company’s own explanation:

Onsemi Convertible senior notes due 2031

onsemi Form 10-Q for the quarter ended 3 July 2026, note 6

The accounting benefits we describe below for net share settled convertibles come at the cost of a binding obligation to fund the principal in cash, which may affect the choice of structure for some companies. However, if the share price has risen far enough to prompt conversion, the company is unlikely to have difficulty refinancing the par repayment. Of course, a cash payment is a possibility for traditional convertibles too, given that the bond must be redeemed at par if no conversion takes place.

The absence of embedded derivative bifurcation under US GAAP

In 2020, the US GAAP accounting for many convertibles was changed to an approach that presents the instrument as a single liability, with no bifurcation of the embedded conversion option. It was this change (effective in 2022) we commented on, and objected to, in our article ‘Convertible accounting: New US GAAP inflates earnings’.

The consequence of not bifurcating is that the reported interest expense is the coupon paid plus the amortisation of the issue costs. For onsemi’s zero-coupon notes this produces an expense in respect of the issuance costs, equivalent to an interest rate of about 0.4%, and nothing else. This is a fraction of what it would have been under a bifurcation model.

If the bifurcation approach were applied, where part of the issue price is attributed to the conversion option, as was generally the case prior to 2022 (and remains so for most issues under IFRS), the bond is initially reported at a lower amount and subsequently accreted to the par value. This produces a normal bond interest cost for this portion of the issue. The value attributed to the conversion option is credited to equity, assuming it meets the definition of an equity derivative.

The cost of the option is observable – the accounts do not report it

What that bifurcated interest cost would have been for onsemi is easily observable because the company has bought a call option on the same terms as the one embedded in the notes. The $352m it paid provides a market-based indication of the value of the matching option embedded in the notes, so the $1,500m raised comprises approximately $1,148m in respect of the bond and $352m in respect of the conversion option. A five-year zero-coupon bond issued at $1,148m and redeemed at $1,500m carries an effective yield of approximately 5.5%.3Our calculation, not a figure reported by onsemi. It assumes that the cost of the purchased call equals the value of the written conversion option, which follows from them having identical terms, and applies the residual to a five-year zero-coupon instrument. That is the true cost of the debt component (before adding the amortisation of issue costs).

Analysis of the onsemi zero-coupon convertible into its components

The Footnotes Analyst

To be clear, we are not suggesting that the artificially low interest cost implies any company has misapplied the standard. The accounting described here is the correct application of US GAAP as it now stands, and our disagreement is with the decision taken by the FASB to shift away from a bifurcation approach, rather than with any issuer’s application of it.

The hedge and warrant overlay

In common with most US convertible issuers, onsemi combines its convertible bond with a separate derivative overlay. This involves purchasing a call option to offset the dilution inherent in the embedded written call, and issuing a separate warrant with a higher strike price.

The structure makes the overall financing package less equity-like and more debt-like. The purchased call exactly offsets the embedded conversion option (if issued on identical terms, which appears to be the case for onsemi), so that the two taken together are equivalent to a straight zero-coupon bond issued at a discount, on which no dilution can arise, because any shares delivered on conversion are matched by shares received under the purchased call. What remains is the warrant with its higher strike price, which is why issuers and their banks describe such structures as achieving a higher effective conversion premium. In the case of onsemi, the conversion price for the convertible itself is $161.30 whereas the warrant exercise price is $211.54, double the share price at the time of issue, so the effective premium becomes 100% rather than 52.5%.

The company has therefore converted an instrument carrying a 52.5% conversion premium into one on which dilution begins only at a 100% premium, at a net cash cost of $70.6m. The same economic position could have been reached directly, by issuing a convertible with a 100% conversion premium in the first place. The option embedded in such an instrument would have been worth $70.6m less, and the company would have had to compensate investors for that difference through a higher coupon, or an issue discount.

The overlay converts an interest cost into a charge direct to equity

Packaged as a single instrument, the identical economic position would produce a real interest charge under the US GAAP single liability model. Achieved through three instruments it produces none, because the $70.6m onsemi paid is charged directly to equity. The difference in reported profit is purely a consequence of the accounting and does not reflect the underlying economics. It is not the accounting for the derivatives that is the problem, but the combination of this with the lack of bifurcation of the convertible. Under IFRS, where bifurcation is applied, the accounting outcome makes the conversion feature much more visible.

If the convertible and the accompanying derivatives are all fairly priced, no party is better off in value terms (although an overlay could arguably create value indirectly if it widens the investor base and improves the pricing of the convertible itself). The only certain value effect of the overlay is negative, in that the banks take a share of these transactions in fees, and a more complex structure increases those fees to the detriment of equity investors. The accounting advantage is cosmetic and not a real source of value, in whatever way it is presented.

The effect on diluted EPS

This is where the new structure gives an additional accounting (but not economic) advantage. We think this accounting advantage is an important reason for the shift to the new structure by US issuers.

The same change in accounting that (mostly) removed the bifurcation of convertibles under US GAAP also required the if-converted diluted EPS method for all convertible instruments, and withdrew the treasury stock method. For a net share settled convertible, it also modified the if-converted method. Because the principal must be paid in cash, and only the conversion spread can be settled in shares, interest expense is not added back to the numerator, and the denominator includes only the incremental shares that would be issued on conversion.4For more about the 2020 change see Deloitte, ‘FASB Simplifies Issuer’s Accounting for Convertible Instruments and Contracts on an Entity’s Own Equity’, Heads Up, August 2020, summarising the amendments to ASC 260.

A conventional structure would add 9.3m shares – this one adds none

The consequence for the onsemi convertible is substantial. A conventional convertible on onsemi’s terms would add the full 9.3m shares issuable on conversion to the diluted denominator, with no offsetting interest to add back because none is paid. Under the structure actually used, the incremental share count for the convertible note is the excess of conversion value over principal divided by the average market price, which is nil while the shares trade below $161.30. The newly issued 2031 notes and warrants therefore contributed nothing to the diluted share count in the second quarter of 2026.5The three instruments are actually tested separately for EPS purposes, which produces an odd result between the two exercise prices. The notes add incremental shares once the share price exceeds $161.30. The purchased hedge is always excluded, because options held by a company over its own shares are treated as anti-dilutive. The warrants enter the calculation only above $211.54. Between those two prices the diluted share count therefore reports dilution that the company has in fact hedged away, while below $161.30 it reports none at all, even though the options have real value. The number is unrepresentative in both directions.

The underlying problem is that EPS dilution is, in effect, based on the intrinsic value of the option. If that is zero, as it will be for onsemi while the share price is below the conversion price, no dilutive effect is reported. The modified if-converted method produces much the same outcome as the treasury stock method it replaced, and it shares the same defect. An option that is out of the money still has value, and the right to demand shares is worth something whether or not it could currently be exercised at a profit. A calculation that recognises only intrinsic value will understate the cost of written calls to existing shareholders, and the effect is only revealed by allowing for the fair value of the embedded option.

This matters more than it once did, simply because a large proportion of this year’s issuance has come from volatile companies, for which the conversion option is at its most valuable and the diluted EPS calculation is therefore at its least informative. For investors, we have long argued that the more reliable analytical approach is to ignore diluted EPS and instead work with enterprise value.6See our article ‘Enterprise value: Our preference for valuation multiples’.

The failure to capture time value in EPS dilution calculations is not confined to US GAAP. IFRS also fails in the same way.7See our article ‘The diluted EPS calculation is 50 years out of date’.

Net share settlement under IFRS

Under IFRS, the same net share settled convertible structure produces the opposite result in terms of accounting favourability. Where the principal must be settled in cash in every circumstance, including on conversion, there is no equity component, and the instrument is not a compound instrument at all. The conversion feature is a derivative that the issuer may settle either in cash or by delivering a variable number of shares, which means it is classified as a liability under IAS 32, and not as an equity instrument.

IFRS 9 requires the embedded derivative to be separated from the debt host, and to be measured at fair value through profit or loss. The alternative is to designate the entire hybrid at fair value through profit or loss, which few issuers would choose. On either approach the reported outcome reverses the US GAAP one. The liability host accretes at the straight debt rate, producing the interest charge that US GAAP omits, and the derivative is remeasured every period, so that a rising share price produces a further loss in profit or loss.

An IFRS reporter therefore has little incentive to adopt the structure that is now popular among US issuers. A conventional convertible under IAS 32 gives an equity component fixed at inception and an interest charge that is higher than the coupon but entirely predictable. The net share settled version gives the same higher interest charge together with earnings volatility that the issuer cannot control.

The same instrument and the same economics – two opposite reporting outcomes

We understand that outside the USA convertible bonds tend to be structured in the conventional manner with physical settlement in shares, and that few issues follow the structure now common in the USA. We think this is a clear instance of accounting driving the structure of financing rather than the financing being determined by what is economically best for the company.

Cost of capital for a convertible

While the accounting inadequacies we describe may give a misleading impression of the cost of convertibles in financial statements, even if you can see through this, you are still faced with the challenge of determining the economic cost itself.

As we noted at the outset, convertibles are commonly described as cheap financing. Indeed, FactSet, writing about the Alibaba issue, says the objective is to “raise capital at a low cost while simultaneously monetising the volatility”.8Tom P. Davis, ‘Why Are Top Companies Rushing to Issue Convertible Bonds?’, FactSet Insight, 1 December 2025. Monetising volatility might sound like a benefit to investors but, in reality, it is selling some of the equity upside that would otherwise accrue to existing shareholders. Monetising volatility creates the low coupon and the conversion premium, but it does not reduce the cost of capital.

A zero coupon prices the option – it does not mean a lower cost of capital

The cost of capital for a convertible is the weighted average of the costs of its two components. The debt component costs what the company would pay on comparable straight debt. The option component has a higher beta than the underlying equity because the option is leveraged exposure to the share price, and therefore normally has a higher required return. The overall cost of the convertible lies between the cost of straight debt and the cost of equity, approaching the latter as the conversion option becomes sufficiently valuable that the convertible behaves almost entirely like equity.

Convertibles are not a cheap form of financing, and their economic cost simply reflects their hybrid financing risk characteristics.9See our article ‘Cost of capital for convertible bonds: Zero coupon ≠ Zero cost’.

Below is our convertible cost of capital model, with the terms of the onsemi convertible (excluding the overlay) included as input assumptions, together with other assumptions such as the equity risk premium and the equity beta of onsemi common stock. Our estimated cost of the convertible (excluding the effect of the derivative overlay) at issue is 9.6%. It will be a little lower today, given that the stock price has fallen and the convertible has consequently become more debt-like.

Interactive model: Convertible bond cost of capital

Loading

The Footnotes Analyst

Note: This model is based on a simplified Black-Scholes option valuation and may not reflect the full complexities of a convertible bond and its pricing. You will find more explanations about the model and the simplifying assumptions we have used in the downloadable version.

DOWNLOAD THIS MODEL

Please enter your email address to receive an excel version of this model

I agree to The Footnotes Analyst Terms of Use and Privacy Policy, and to receive email notification of future articles.

You can unsubscribe any time.

The model derives the debt and option components of a convertible from the terms, calculates a cost of capital for each, and shows how the blended cost changes as the security becomes more bond-like or more equity-like as the share price moves. You can vary the input assumptions to produce your own estimate. The model is free to download.

Insights for investors

  • Do not read a low reported interest charge as evidence of cheap financing. Under US GAAP a zero-coupon convertible carries almost no interest expense, but the cost has been paid by writing the call option.
  • Where a company has purchased a hedge on the same terms as the conversion option, the price it paid provides a market valuation of the embedded option and, by subtraction, the implied straight debt yield.
  • Treat the net cost of a hedge and warrant overlay as interest. It replaces the higher coupon the company would have had to offer on an equivalent convertible, and it is charged to equity rather than to profit.
  • Where the principal must be settled in cash, diluted EPS may record no dilution at all. The denominator captures only the excess of conversion value over principal, so a conversion option that is out of the money adds nothing, however valuable it is.
  • Adjust for the settlement structure before comparing a US issuer with an IFRS reporter. The same instrument produces a lower interest charge and a smaller added diluted share count under US GAAP than it would under IFRS.
  • Prefer enterprise value, with the convertible included at fair value as a claim on the enterprise. This avoids the accounting deficiencies of US GAAP, the comparability issue with IFRS, and the limitations of the EPS calculation under both.

Don’t miss new articles published by The Footnotes Analyst – subscribe here:

none
Loading

No advertising, no spam, just new articles. Unsubscribe anytime.

Send us a question or comment about this article:

    For related articles select subject tags or see the suggestions below:

    Print or save article as a pdf: