Amazon free cash flow – an update

Last year we published an article about the calculation of free cash flow and the alternative approaches used by Amazon. That original article is still very relevant; recent accounting changes have prompted us to publish an update.

New accounting rules effective in 2019 change and improve the data available to you when making the adjustments we advocate. We explain these changes, provide updated free cash flow measures for Amazon based upon their 2019 financial statements, and consider the relevance of maintenance and growth capex in the analysis of free cash flow.

Click here to read our earlier article – ‘In search of free cash flow – Amazon’

Free cash flow is an important metric in equity analysis and valuation. It is used in cash conversion metrics, in valuation multiples and it forms the basis for most discounted cash flow valuations. But there is no standardised definition of free cash flow in accounting standards and you will find different versions presented by companies.

The most common calculation of free cash flow is operating cash flow, as reported in the cash flow statement, less capital expenditure. You may additionally see the deduction of lease payments, particularly for IFRS companies, following the changes made to lease accounting. However, we argue that both of these approaches are incomplete and not suitable for use in equity valuation.

Amazon makes an adjustment in its their free cash flow calculations that most companies ignore

The problem concerns certain, seemingly non-cash, transactions which in economic terms involve equal and offsetting flows. Leasing is a good example. At the inception of a lease contract there is no actual cashflow; however, there is an inflow of finance and an outflow in respect of the purchase of a right-of-use asset. We argue that because the outflow is, in effect, capital expenditure then it should be included in free cash flow metrics.

What attracted us to Amazon for this analysis is that this company already includes the adjustment we advocate in their own free cash flow calculations, disclosed in their management commentary. The company provides three alternative versions of free cash flow, the first being the traditional approach and the second with adjustment to include lease payments. However, the third, which we illustrate below, includes the deduction of the effective cash flow “Equipment acquired under finance leases”1In prior years Amazon used the term capital leases for finance leases of equipment. We also used the term capital leases in our previous article., exactly as we suggest.

Amazon free cash flow calculation (Version 3)
Source: Amazon 2019 financial statement

We commend Amazon for their recognition of the capital expenditure related to finance leases of equipment; however, there are three further effective flows that are not reflected in their measure, but which in our view should be.

Our additional adjustments are:

  • Finance leases of property and other asset financing arrangements: Amazon does not apply their equipment finance lease adjustment to similar leases of properties. We see no reason not to. Purchases of property, whether directly or through lease transactions, are just as much capital expenditure and core to the business as are purchases of equipment.
  • Operating leases: Amazon also omits all assets acquired as a result of operating leases. This is complicated by the change in accounting in 2019 and also by the specifics of US GAAP (IFRS is different). In our view ‘right-of-use’ assets acquired through operating leases should also be part of capital expenditure.
  • Share-based payments: Non-cash share-based payments are, in effect, two offsetting flows – a remuneration outflow to employees and a financing inflow whereby these employees are providing equity finance to the business. We think that the effective operating outflow should also be part of free cash flow calculations.

New accounting and new data

The major difference between the 2018 and 2019 financial statements of Amazon is that operating lease liabilities and the related right-of-use assets now appear on the balance sheet.

Prior to 2019, operating lease payments were recognised as an expense with no asset or liability included in the balance sheet. This meant that we previously had to estimate the liability using disclosures about lease commitments to then derive an estimate of the operating lease capital expenditure from changes in this liability.

Amazon disclosure of their 2019 capitalised operating lease liability
Source: Amazon 2019 financial statements

The new disclosures help a lot but there is still an element of approximation required for US GAAP reporters. Amazon discloses the new operating lease capital expenditure immediately below its cash flow statement – the figure is $7,870m. However, due to the US GAAP ‘single lease expense’ approach to operating lease accounting, there is no disclosure of the split of lease payments between interest and principal (debt repayment) amounts. We need this because the principal component2The Amazon free cash flow measure is an ‘equity’ flow and therefore stated after deducting interest expense. Free cash flow measures can also be ‘enterprise’ flows, which are pre-interest. If that were the case our adjustment would be easier because we would have added back the full operating lease payment. must to be added back in our calculation to avoid double counting. Nevertheless, the estimation is relatively easy. Multiplying the disclosed average discount rate by the average lease liability should give you a good approximation of the interest accretion. The amount of principal repayment is then the difference between this interest and the total operating lease expense.

Under IFRS there is no distinction between operating and finance leases and the interest accretion and principal repayment amounts are always presented separately. To learn more about the difference between IFRS and US GAAP lease accounting, including how they are presented differently in the cash flow statement, see our article ‘Operating leases: You may still need to adjust’.

Amazon disclosure of the ‘purchase’ of operating lease right of use assets
Source: Amazon 2019 financial statement. This disclosure appears immediately below the cash flow statement. The columns are 2017 to 2019 left to right.

The 2019 data provided by Amazon regarding operating leases has enabled us to refine our previous estimates. We now believe that we previously underestimated the 2018 operating lease liability due to a combination of using too high a discount rate and too short a lease duration. Fortunately, our two estimation errors partially cancelled out and overall, we think we were quite close, but this illustrates the benefit of the new lease accounting rules for investors who no longer have to make what may often have been very rough approximations.

Amazon free cash flow updated

The table below reproduces the 2016 to 2018 data that formed the basis of our previous article, supplemented by the new data disclosed by Amazon and our alternative free cash flow estimate for 2019. We have not adjusted our previous estimates of The Footnotes Analyst measure of free cash flow for 2016 to 2018. A refined estimate, taking into account the new information available in 2019, would have been lower, but not significantly so.

Summary of Amazon’s free cash flow metrics and our preferred approach
Source: Amazon 2019 annual report and The Footnotes Analyst estimates. In 2019 Amazon changed the terminology it uses for leases – the table reflects that new terminology.

One difficulty in determining the operating lease capital expenditure arises where a company has made significant acquisitions during a period and the acquired companies have operating leases. Capital expenditure in free cash flow calculations should not include the newly consolidated right-of-use assets of acquired companies. This is not a problem in 2019 because of the new accounting. The amount of $7,870m disclosed by Amazon for ‘Assets acquired under operating leases’ would automatically exclude the effect of acquisitions. However, for our estimated figures in earlier years we would need to separately identify and exclude the relevant amounts. This is difficult in practice because it requires examining the individual financial statements of companies acquired.

Acquisitions make it difficult to compute operating lease capex

The most significant business combination effect for Amazon is their acquisition of Whole Foods Market in 2017. We estimate that at the time of acquisition Whole Foods had an operating lease liability of over $6bn. Had we not adjusted for this when deriving the operating lease liability roll forward, we would have overstated the operating lease capex and, hence, understated our measure of free cash flow. Our figures above have been adjusted to remove the (estimated) effect of this acquisition, but we have not made any attempt to adjust for other acquisitions.

From 2019 you should find that the operating lease capital expenditure is disclosed, but the lack of a requirement to present a detailed lease liability roll forward in financial statements may still make it difficult to identify the effect of business combinations.

Maintenance versus growth capex

Although our revised free cash flow metric is much lower than those presented by Amazon, we do not believe this is necessarily a negative indicator regarding Amazon as an investment. Negative free cash flow can arise for two reasons, either a company is simply not generating any cash due to poor performance and underlying low profitability or because the cash it is generating is being reinvested to sustain a high rate of growth.

High growth makes the analysis of capex into maintenance and growth components more important

In the case of Amazon, high growth is clearly evident. Revenue growth in 2019 was 20.5% and the average annual compound revenue growth since 2015 has been 27.2%p.a. Some of this is due to acquisitions, but most is organic. Considering the high growth, it is of little surprise that free cash flow, after allowing for all capital expenditure, is low or negative.

In these circumstances it may be worthwhile to attempt to split capital expenditure into ‘maintenance’ and ‘growth’ components and present an additional free cash flow metric before deducting growth capex. Maintenance capex is not disclosed in financial statements. Some companies provide a voluntary disclosure, but this is not common. As a result, you will need to estimate.

Probably the best place to start is the depreciation and amortisation expense because this represents an allocation of assets already held. However, you should consider two adjustments:

  • Price and technology changes: Depreciation reflects the cost of assets currently held. If the same asset would not be purchased today or, at the same price, then maintenance capex may be higher or lower.
  • Intangible amortisation: Some amortisation relates to intangibles that are unlikely to be explicitly replaced by further asset purchases, but that are ‘maintained’ through expenditure that is itself expensed. This often arises in respect of intangibles capitalised as a result of past business combinations where, had the asset been internally developed, it would not have been recognised. Many companies adjust for this component of amortisation in their alternative performance measures and is a subject we consider in our article ‘Should you ignore intangible amortisation – AstraZeneca’.

Our estimated split of the Amazon capex into maintenance and growth components reflects the depreciation and amortisation of Amazon, with two adjustments. First, we estimate the depreciation applicable to right-of-use assets arising from operating leases. Under IFRS this would not be a problem but due to the US GAAP ‘single lease expense approach’ the depreciation amount is not provided. Secondly, we exclude intangible amortisation for the reasons we explain above, although we have not attempted to separate the amortisation which may be indicative of maintenance capex.

The Footnotes Analyst version of Amazon free cash flow – maintenance v’s growth capex
Source: Amazon financial statements and The Footnotes Analyst estimates

Although estimating maintenance capital expenditure is challenging, we think it is worth the effort to do the analysis. It forces consideration of what level of additional investment is driving growth, which helps in assessing the incremental returns on that investment. We also think that free cash flow, after deducting only maintenance capex, is a better basis for price (or EV) to free cash flow multiples.

Insights for investors

  • Free cash flow calculated by deducting capital expenditure from operating cash flow misses important ‘effective’ cash flows. Using this metric in, for example, DCF analysis is likely to give erroneous results.
  • Adjust free cash flow measures to reflect all capital expenditure, including that arising from new lease transactions.
  • New accounting and disclosures under IFRS and US GAAP make this leasing adjustment to free cash flow much easier from 2019.
  • Remember that ‘effective’ flows can arise from other transactions such as stock-based compensation and supply chain finance.
  • Free cash flow, particularly for high growth companies, may be best analysed by separating growth capex from maintenance capex.

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


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: