EV to EBITDA multiples must be consistent

Swiss pharma company Novartis provides investors with its own calculation of an EV/EBITDA multiple. However, in our view, the EV is inconsistent with EBITDA. We review the company’s calculation and suggest amendments to ensure it better captures the value of Novartis’ core business.

To derive useful valuation multiples, you must be consistent. Our main adjustment to the Novartis calculation relates to the value of their stake in fellow Swiss pharma company Roche.

In our article, Enterprise value – calculation and miscalculation, we identify practical challenges and common mistakes in the calculation of enterprise value and related valuation multiples. The calculation of an EV/EBITDA multiple by Novartis, and its disclosure in the company’s 2018 annual report, is welcome, but it also provides a good illustration of these challenges.

Novartis refers to enterprise value as the total amount that shareholders and debt holders have invested in Novartis, less the Group’s liquidity. In contrast, we view enterprise value as the current value of all financing claims including equity, (net) debt and other debt-like liabilities. The Novartis approach is a combination of a market value for one claim (market capitalisation) and the book value of investment made by other providers of capital. We think this is inconsistent and incomplete.

Novartis EV/EBITDA calculation

We do not examine the merits or otherwise of using EBITDA as a basis for valuation multiples – we have our reservations. Our focus is on Novartis’ calculation of enterprise value and whether it is complete and consistent with the EBITDA metric they use in the multiple.

Calculating enterprise value for use in valuation multiples requires identifying the various financing claims on that business. By ‘claims’ we mean the interest of debt holders, shareholders and others who benefit from that business enterprise through a return on their investment. As we discussed in our previous article, it is important to distinguish between total and operating enterprise value. Total enterprise value is the sum of the value of the company’s operating activities plus any other value creating activities, such as investments in associates not treated as part of operations.

Many companies including Novartis separately identify an operating profit subtotal in their income statement. Operating enterprise value is the value related to activities that contribute to operating profit and equals total enterprise value minus the value of any business activities that do not contribute to operating profit.

If you are given an enterprise value or EV multiple there are three key questions you should ask: Is it consistent, is it complete, and does it reflect current values? We examine each in the context of the calculation by Novartis.

Is enterprise value consistent with the related profit?

In using an EV/EBITDA multiple, it is necessary to be consistent between the definition of enterprise value and the related EBITDA number. It is crucial that EBITDA is related to the value of those financing claims that contribute to, and receive benefit from, this EBITDA.

This is an important issue for Novartis. The company owns a 33.3% stake in the voting shares of the other Swiss pharmaceutical company Roche. This represents approximately 6.2% of Roche’s total outstanding voting and non-voting equity instruments. Novartis applies equity accounting, meaning that its balance sheet reflects Novartis’ share of Roche’s net assets plus acquisition goodwill (investment in associates) and reports its share of Roche’s earnings in its income statement (income from associates). The latter is presented below operating profit.

Novartis disclosure of investment in Roche

Novartis shareholders clearly benefit from the stake in Roche, which is therefore implicitly reflected in the Novartis market capitalisation. However, EBITDA, as defined by Novartis, does not include the earnings contribution. We believe this results in an inconsistent multiple.

There are three potential ways to adjust the multiple in respect of the investment in Roche.

• Add the earnings contribution from Roche to Novartis’ EBITDA: The problem is that the multiple is then a combination of an EV/EBITDA multiple for Novartis’ operating activities and, in effect, a price earnings ratio for Novartis’ share of Roche’s net earnings. The resulting mixed multiple would be impossible to interpret.

• Adjust Novartis EBITDA to include its share of Roche’s EBITDA: In this case the Novartis enterprise value would need to be adjusted to include the share of Roche’s enterprise value, excluding the market value of its equity investment, as this is already captured in Novartis’ current enterprise value. Although this produces a more consistent multiple, it needs to include the proportionate share of the different components of Roche’s enterprise value and a proportionate share of its EBITDA, making it challenging from a practical perspective.

• Remove the market value of the investment in Roche: This approach produces an EV/EBITDA multiple that relates the implied value of Novartis’ operating activities to the EBITDA generated by those operating activities.

We prefer the third approach as a more practical way to get a consistent multiple, where the value of Novartis’ operating activities is linked to the group operating results.

Novartis also reports financial assets of US$2,345m in the balance sheet. The income related to this appears to be in ‘other financial income and expense’ in the income statement, which is not included in the EBITDA used in their valuation multiple. This a further inconsistency which we adjust for in calculating our revised operating enterprise value and EV/EBITDA multiple.

Inconsistencies can also arise from the treatment of liabilities and their related financing expense. Some liabilities can either be regarded as operating or financing for the purpose of valuation. In many cases the nature of the liability is clear, but for other liabilities valuation multiples can be successfully calculated in different ways. However, it is important to be consistent because treating a liability as operating means the financing element (interest accretion) should be treated as an operating expense. 

For Novartis this issue potentially applies to its pension and environmental liabilities. However, there is no inconsistency in their calculation as the company has not included these claims in enterprise value and the interest accretion is treated as an operating item in profit and loss.

Our preference would be to treat these liabilities as financing, for reasons we outline below.

Is enterprise value complete?

It is important that enterprise value reflects the value of all financing related claims on the underlying business. We believe a number of additional items could be included in the Novartis enterprise value.

• Pension and environmental liabilities: Novartis is consistent in its treatment of pension and environmental liabilities. However, we prefer treating these as financing. They are long term, measured on a present value basis and hence incur an interest accretion expense. Missing out these claims potentially reduces comparability with other companies. For example, one company may have a pension deficit whereas another company may have raised additional capital to add to the pension fund and eliminate its deficit. Treating the pension deficit as operating would potentially produce different multiples that reflect differences in capital structure rather than underlying differences in value.

• Operating leases: We believe that operating leases are a form of financing and, as such, should be capitalised and included in enterprise value. Of course, we would therefore need to adjust EBITDA for the depreciation of the lease asset and the interest accretion on the capitalised lease liability.

In our adjusted multiple below we take the EBITDA presented by Novartis as our reference point. Consequently, for the purpose of this calculation, we have not included the pension and environmental liabilities as part of enterprise value and, similarly, have not made any adjustment in respect of operating leases.

However, we do adjust for the following:

• Contingent consideration: Novartis presents liabilities that reflect future payments to subsidiary shareholders. These give rise to an interest accretion expense.  Given the treatment of the interest expense as a financing item (below operating profit), we believe the liability for contingent consideration of US$907m should be included in the calculation of enterprise value with no adjustment to EBITDA.

• Share repurchase commitments: This is a commitment to repurchase shares that reflects an arrangement with a bank to buy shares for US$284m. While this appears to be a financing liability, it is important not to double count because the share repurchase would reduce market capitalisation. The liability should only be included in EV if the share count used for market capitalisation has already been reduced by the number of shares to be repurchased. When we queried this with Novartis, the company stated that “these shares are therefore not included in Novartis current number of shares”. Accordingly, we have reflected this liability in our enterprise value with no adjustment to EBITDA.

• Employee shares and options: Novartis currently has 25.7 million unvested shares and a further 5.6m equity options outstanding that arise from equity-based participation plans for employees. In our view, the market value of the shares that are expected to vest (after allowing for expected forfeits) and the value of the options should both be included in enterprise value. As there is no guidance on expected forfeiture of these unvested shares, we include the fair value of all unvested shares (US$2,190m). The option programme was discontinued in 2013, but there remain 5.6m options outstanding with an average exercise price of US$59.9 compared with a year-end share price of US$85.0. Given that the options are significantly in the money, and that consequently time value would be small, we include the current intrinsic value of the options of US$142m in our enterprise value.

Does enterprise value reflect market value rather than book value?

In deriving any valuation multiple, the market value of the claim is related to the earnings for the claimholder.  This is no different for enterprise value multiples. Novartis, therefore, rightly includes the market capitalisation rather than the book value of equity in its calculation of enterprise value. Market value should also apply to all other claims, including non-controlling interest (NCI), as this is just another source of equity capital. Novartis has included the book value of NCI in its EV/EBITDA multiple.

Where the non-controlling interest relates to a subsidiary that is quoted it should be possible to directly measure this claim at fair value. However, this is often not the case. One may need to estimate a value based upon available information for the subsidiaries concerned. Information about NCI in consolidated financial statements is generally limited and often the only way to value this claim directly is to apply an appropriate PE ratio to the NCI share of net income or a price to book ratio to the book value of NCI. In most cases this would be sufficient.  However, if NCI is highly significant it is worth further analysis.

The value of NCI in the case of Novartis is small and even ignoring the claim entirely does not materially affect enterprise value or related multiples. Nevertheless, we include an adjustment to reflect estimated NCI market value for the sake of completeness.

In the case of interest-bearing debt, there is generally only a material difference if the company has issued fixed rate debt and there is either a change in interest rates or a change in creditworthiness of the company. In the case of Novartis, the market value of interest-bearing debt at Novartis is broadly the same as its book value.

Our calculation of EV/EBITDA

Not all of our adjustments to the EV/EBITDA multiple of Novartis have a material impact and others tend to offset each other leaving the overall multiple lower by only a modest amount.

However, we consider the adjustment in respect of the investment in Roche to be particularly important and it is the primary reason why Novartis attracted our attention. Without this adjustment we think the EV/EBITDA multiple for Novartis is overstated. We also believe that the inconsistency between the measurement enterprise value and the related profit measure, such as EBITDA, is a common mistake in practice, including by the various data providers.

Revised EV/EBITDA multiple for Novartis

Insights for investors

Check enterprise value and the related valuation multiples that you use, or calculate, are complete and consistent. Ask the following:

• Is enterprise value consistent with the related performance measure?

• Is enterprise value complete?

• Are the components of enterprise value measured at market value and not book value?