A forecast of profit is used for both valuation multiples and as a starting point in deriving free cash flow for DCF valuations. But should you use a forecast of the reported IFRS or GAAP measure, or a forecast of the adjusted non-IFRS or non-GAAP alternative performance measure (APM) presented by management?
We think equity valuations should be based on forecasts of reported IFRS or GAAP earnings (albeit with some adjustment related to intangible assets). Forecasts of management APMs can be useful for understanding trends in performance but using these in equity valuation is likely to introduce a structural bias.
In deriving valuation multiples, we have previously discussed the relative merits of different measures of profit, such as earnings and EBITDA, and the importance of using forecasts, particularly when part of a more comprehensive forward priced multiple.1For more about why you should avoid EBITDA see our article ‘EBITDA-AL: More letters but no more insight’ and for more about forward priced multiples go to ‘Why you should forward price valuation multiples’. But there is a further choice to make – whether to use a forecast of the reported measure as presented in financial statements or the adjusted (non-IFRS or non-GAAP) measure presented by management.
Considering that management guidance invariably focuses on the adjusted metrics, and that generally these are easier to model, most forecasts in practice – including the often-quoted consensus amounts – seem to be forecasts of non-IFRS or non-GAAP management metrics. However, we do not think using these gives the most useful valuation multiples. In our view, you should always also forecast IFRS (GAAP) earnings and it is preferable to use these metrics for valuation multiples (and as the starting point for deriving free cash flow in DCF analysis), albeit with adjustments to remove the mismatch resulting from the accounting for certain intangible assets.
APMs and analysing performance
Clearly, forecasting IFRS or GAAP performance metrics is challenging. After all, one of the main objectives of adjusted non-IFRS earnings is to provide a more stable metric that gives a better indication of underlying trends in profitability. This is achieved by excluding certain volatile or one-off items, most commonly things such as restructuring costs, asset impairments, unusual effects of tax rate changes or fair value changes for items measured at fair value through profit and loss.
Global pharmaceutical company GSK includes period to period comparisons as a reason to use such metrics. Below we use their non-IFRS disclosures to illustrate different non-IFRS adjustments.
“GSK believes that Adjusted results, when considered together with Total results, provide investors, analysts and other stakeholders with helpful complementary information to understand better the financial performance and position of the Group from period to period, and allow the Group’s performance to be more easily compared against the majority of its peer companies. These measures are also used by management for planning and reporting purposes.”
GSK financial statements 2021
There is nothing wrong with reporting performance before deducting these items as long as the nature of the measure, and what is omitted, is clearly explained. Presenting the APM and the excluded items in the form of a disaggregation of the IFRS result provides investors with better information than only reporting the IFRS compliant profit. This is particularly the case when the disaggregation provides the opportunity to separately analyse components of performance with differing characteristics, such as value changes versus flows, or differences in the degree of persistency.
The lack of sufficient disaggregation of performance measures in practice has been a recurring criticism by us of financial reporting. For example, in ‘Forecasting sticky stock-based compensation’ we highlight the problem investors face in interpreting the stock option expense, where this is affected by catch-up adjustments that are not disaggregated from the underlying or current period component of the expense. In ‘Disaggregation is key to understanding performance’ we discuss proposals by the IASB to improve the transparency of performance metrics, which includes further disaggregation requirements. More Footnotes Analyst articles that include commentary on different aspects of disaggregation can be found here.
However, while using APMs can help to identify trends in ‘underlying’ profitability, we think APMs are not suitable as the basis for valuation multiples.
APMs and valuation
The problem with using forecasts of adjusted profit metrics in valuation multiples or as a starting point in DCF, is that the adjustments applied in APMs do not necessarily have a zero expected value in future periods.
Take the commonly excluded restructuring costs: even though this might be a volatile item and a particular restructuring may be a one-off event, it is difficult to see how the expected cost of restructuring would be zero in all future periods. Using a valuation multiple based on the pre-restructuring cost profit ignores this expected future cost. Of course, if multiples of other companies are calculated on the same non-GAAP basis, the analysis may be more valid, but this assumes that the excluded items are consistent across companies, which is unlikely.
It is not just the so-called non-recurring items that are a problem, but also the commonly excluded value changes. An item we often see excluded from APMs is the change in value of deferred consideration in a business combination. Deferred consideration liabilities are measured at fair value (the present value of the expected settlement amount) at the time of a business combination and updated each subsequent period until they are settled. Any change in fair value is reported in net income.
Clearly the changes in the fair value of deferred consideration are likely to be volatile. We have no objection to the use of APMs that exclude this amount to analyse trends in performance. The overall future change in liability for deferred consideration will be impossible to forecast because it reflects changes in cash flow expectations and the time value of money. However, this does not make the best estimate of the gain or loss zero. The fact that the liability is stated at a present value means a discount unwind is built into the valuation and, even if cash flow expectations and interest rates do not change, there would be an expense in future periods as a result of this interest accretion.
APMs tend to be greater than the equivalent reported GAAP or IFRS amount. This is simply because more of the adjustments are usually expenses rather than income, such as our two examples above, and is partly a function of the conservative nature of accounting. Using forecasts of APMs as a basis for valuation creates the risk of systematically overstating forecast profits and hence current value.
This structural bias is most obvious for so-called absolute approaches to value, such as DCF, where starting with an artificially inflated profit measure is likely to lead to unrealistically high values. For relative valuations, such as valuation multiples, the higher profit measure could lead to lower multiples, suggesting the company in question is undervalued compared to its peer group. In both absolute and relative valuation, the valuation conclusion might be that the company is valued attractively whereas this is nothing more than a consequence of using the inflated forecast earnings number.
However, we do not suggest using forecasts of IFRS / GAAP profit without any adjustment. Certain intangible amortisation and impairments, and gains or losses related to business disposals, should be excluded from metrics used for equity valuation, even if forecasts of these items are non-zero. We explain why below.
Forecasting non-GAAP adjustments and what to do with them
To forecast GAAP or IFRS earnings it is necessary to forecast the adjustments made by management, in addition to the non-GAAP metric itself. These forecasts depend on whether your focus is on earnings or metrics related to enterprise value, such as EBIT or NOPAT. For EV based analysis, only the adjustments that relate to operating activities will need to be forecast. For example, in the case of GSK below, one of the prominent earnings adjustments concerns the value change reported for deferred and contingent consideration for past business combinations. If this liability is treated as a ‘debt equivalent’ and therefore a component of enterprise value (which we think it should), the change in that liability including the interest accretion, would not be relevant, whereas in an earnings-based analysis it would.
To help explain what we think you should do with the adjustments made by companies we use the non-IFRS disclosures of GSK.
GSK 2021 Adjusted results reconciliation
GSK 2021 financial statements. Additional explanations and disaggregation are provided in supporting notes to this table – see pages 70 to 72 of the 2021 annual report.
GSK provides good disclosures and explanations of their adjusted non-IFRS performance measures. However, we think it is useful to classify these adjustments based on a combination of whether forecasts are likely to be non-zero, how they might be forecast and the extent to which these forecasts are relevant for equity valuation.
Here is our classification of the adjustments applicable to profit attributable to the shareholders of GSK (the metric that would be used in a price earnings ratio) for the last 8 years. You will notice that our analysis includes additional disaggregation compared with the table above; this is based on the supporting disclosures provided by the company.
Analysis and classification of GSK adjusted earnings
GSK 2014-2021 adjusted results reconciliations and related disclosures
Recurring but volatile – non-zero expected value
Some items are excluded from management performance metrics simply because they are volatile, with profit measured after including the gain or loss being less useful in identifying trends in performance. However, these items almost certainly have a non-zero expected value, will impact future cash generation and hence affect value. We think investors should forecast these items and include them in the metrics used as a basis for valuation.
Forecasting will be difficult and subject to a wider margin of error than other components of performance. You may find that the best you can do is to use an average of historical reported amounts. However, we think this is better than using zero, which will inevitably be incorrect.
In the case of GSK, we have put in this category their adjustments to exclude major restructuring costs and certain legal expenses. The restructuring expense, in particular, clearly does not have a zero expected value and is material to any valuation.
Value changes with expected accretion or return
It is common to see companies exclude certain value changes from their adjusted performance metrics. These could be investments reported at fair value through profit and loss, derivatives for which hedge accounting does not apply or is not available, or liabilities that are reported at a current value where interest rate changes create gains and losses in profit and loss. Such adjustments are understandable, particularly for ‘non-financial’ sectors, because it is difficult to interpret profit and establish trends when it is a combination of ‘flows’, such as operating revenue and expenses, and value changes.
However, excluding the value changes means that earnings measures do not include any benefit from these assets, or any cost related to liabilities. The expected future return or cost for these items will always be non-zero simply due to the unwinding of the discount inherent in the current valuation. Although it is almost certainly impossible to forecast actual realised value changes, we think that the discount unwind or expected return or cost should be included in forecast earnings.
For GSK we include their remeasurement of deferred consideration and NCI puts2NCI puts arise where the minority or non-controlling shareholders in a subsidiary have an option to sell their holding to the parent company. The NCI put is reported as a liability in the consolidated financial statements equal to the present value of the exercise price of the put option, which is updated each period. arising from business combinations in this category. These are significant liabilities for GSK and, considering the risk adjusted discount rate that is applied, the expected accretion has a material impact on forecast IFRS earnings.
Recurring but excluded by management for other (often spurious) reasons
Some items (generally expenses) may be excluded from adjusted performance metrics even though they are not volatile and do not fit in any one of the categories above. Some companies use the argument that an expense is non-cash, such as the expense for share settled equity-based compensation.
Other companies may argue that a particular expense is outside of management control. While these reasons may have some relevance for internal management reporting we do not think they are a reason for investors to exclude gains or losses when both analysing performance and for equity valuation. While forecasting stock-based compensation can be challenging, we think it is important to include this expense in all performance metrics.3Although we think that stock-based compensation must be included in any performance metric used for analysis and valuation, we recognise that some components of this expense may be non-recurring. For a discussion about forecasting the SBC expense see our article ‘Forecasting sticky stock-based compensation’.
Apart from the business disposals and intangible amortisation expense we discuss below, we think that any gain or loss with a non-zero expected value should be included in forecast metrics used for valuation.
We note that GSK does not have any adjustments which we would put in this category.
Non-recurring or impossible to forecast – zero expected value
There are some gains and losses that truly are non-recurring or that, while they may recur in the future, the best estimate of the future value is zero – usually because there is an equal chance of there being a gain or loss – or the item is simply impossible to forecast. For these items include zero in a forecast of IFRS earnings.
For GSK we have put their tax related items in this category, including adjustments to deferred tax due to tax rate changes. Deferred tax assets and liabilities are based on current and expected tax rates and if that rate changes (or is reasonably certain to change in future periods) deferred tax is adjusted with the gain or loss reported as part of the tax charge in profit and loss. While future adjustments are likely, it is impossible to identify whether these will be gains or losses, let alone their magnitude.
When companies dispose of a subsidiary or business unit, a gain or loss arises equal to the difference between the disposal proceeds and the net assets of the business that are removed from the consolidated financial statements, including any goodwill that was recognised if that business was previously acquired. For companies that trade in businesses regularly, including GSK, these gains and losses can be a recurring source of profit. Even though the amount is difficult to predict, the best estimate for a forecast may not be zero.
However, we do not think that forecast gains from business disposals are relevant from a valuation perspective. On average, the effect of a disposal is more likely to be a gain simply due to the inherent conservatism of accounting in not recognising most intangible assets that are created after a business has been acquired. In part, business disposal gains offset the double counting that we explain below in respect of acquired intangibles, and should similarly be excluded from profit forecasts, even if expected to be non-zero in future periods.
Assume a business is bought for 100 when net tangible assets are 40, the acquired intangibles are valued at 50 and therefore goodwill is 10. Assume that 5 years later the business and its value have not changed, and it is sold for the same 100. During the 5 years the acquired intangibles have been amortised to zero, but replacement intangibles have not been capitalised as they do not qualify for recognition under (conservative) accounting rules. The gain on sale will be 50 (100 – 40 -10), which exactly offsets the (double-counted) intangible amortisation recognised in prior periods. The gain on sale is not an economic gain as the sale proceeds is the same as the purchase price. Equally the decline in intangible value is not an economic loss – the intangibles have been replaced but just not recognised as an asset.
The expectation of gains is illustrated by the fact that GSK shows business disposal gains in each of the last 8 years, even after allowing for the separation costs which we consider linked with business disposals.
In addition, if you include a forecast of business disposal gains in metrics used for valuation, it means that value is likely counted twice – forecasts of group earnings would include (1) earnings of all current businesses, including those (as yet unidentified) subsidiaries assumed to be sold, and (2) the assumed gain on the disposal of said subsidiaries.
Intangible amortisation and impairment
The inconsistent accounting for intangible assets results in perhaps the most common of the non-IFRS or non-GAAP adjustments – the exclusion of intangible amortisation and impairments related to intangibles acquired in a business combination. Because most internally generated intangibles are not recognised there is a form of double counting related to ‘replacement-expensed’ intangibles acquired in a business combination. This arises because the amortisation of acquired intangibles appears in the same period as the expenditure that contributes to their enhancement or replacement.
We believe reversing the amortization charge for purchased intangibles makes sense where subsequent expenditure to enhance or replace the intangibles in future periods is expensed and not capitalised, such as for marketing related intangibles, including customer lists. We do not believe the add back of intangible amortisation is justified for ‘replacement-capitalised’ intangibles – those where expenditure on replacement assets is capitalised. In this case, we would classify any add back of amortisation into the ‘recurring but excluded by management for other (often spurious) reasons’ category described above.
One solution to the double counting problem for replacement-expensed intangibles is for investors to adjust the recognition of intangibles to better reflect the economic investment – in other words applying estimated adjustments to capitalise the ‘future-orientated expenses.. But, in practice, often the only realistic approach to dealing with this double counting is to add back the amortisation and impairment of ‘replacement-expensed’ intangibles – generally most of the intangibles recognised in a business combination.
Even though replacement-expensed intangible amortisation may be a recurring item and relatively easy to forecast, we do not think that you should include it in forecast metrics used in equity valuation.
Identifying whether expenditure on replacement intangible assets is capitalised or not can be difficult. For example, the business model of pharmaceutical companies may involve acquiring intellectual property through separate asset purchases or further business combinations. For GSK we have classified all the intangible amortisation the company adds back in its non-GAAP earnings as ‘replacement-expensed’. If separate acquisitions of intellectual property occur frequently, the amortization charge for at least some of the intangibles acquired in a business combination can be seen as no different from other depreciation and amortisation. Reversing out the full intangible amortization charge may then not be wholly appropriate.
For more about the effects of intangible asset accounting on equity analysis, see our previous articles – Should you ignore intangible amortisation; Missing intangibles distorts return on capital; and Intangible asset accounting and the value false negative.
Summary of valuation implications of non-IFRS or non-GAAP adjustments
Insights for investors
- Adjusted (non-IFRS or non-GAAP) performance metrics are relevant in analysing trends but may not be relevant for use in valuation multiples and DCF.
- Many items excluded from adjusted metrics would likely have a non-zero expected value in future periods and should be included in forecasts of IFRS and GAAP earnings and, if operating in nature, in operating profit.
- Always include in your forecasts expected values for restructuring costs and similar volatile items, even if these forecasts are subjective and companies do not provide guidance.
- Always include in your forecasts the expected return or discount unwind for items reported at current value.
- Never use metrics in valuation that exclude stock-based compensation or other items management ignore due to being supposedly non-cash, outside management control or just because their peers do so.
- Business disposal gains and (replacement-expensed) intangible amortisation and impairment may have a non-zero expected value. However, these items are unlikely to be relevant for valuation due to inconsistencies in intangible asset recognition in financial statements.