Amazon provides investors with three alternative calculations of a free cash flow metric. For 2018 these range from $8.4bn to $19.4bn. In contrast our preferred approach gives a negative free cash flow of $3.4bn. What explains these material differences?
The disclosures by Amazon about its free cash flow measures are good and the calculations go further than many other companies. However, in our view important components are missing. We explain our additional adjustments in respect of leased assets and stock-based compensation.
In May 2020 we published an update to this article in which we explain how new lease accounting rules affect the free cash flow adjustments we advocate. We also provide additional analysis of alternative free cash flow approaches and update the Amazon data to include 2019.
Click here to view the update
In a recent article we highlighted the problem of cash flow subtotals, such as free cash flow, not being complete unless certain non-cash transactions are included. These are transactions where there are, in effect, two offsetting flows. The example we used was capitalised leases, where at the inception of the lease there is typically no actual cash flow, but where an effective finance inflow offsets an effective capital expenditure outflow.
The online retail giant Amazon makes significant use of leasing for its expansion and provides a good illustration of the problem faced by investors. The company itself recognises the issue and provides useful disclosures on the face of the cash flow statement. It also includes alternative (non-GAAP) free cash flow measures in its management commentary, one of which reflects these effective cash flows.
This extract from the Amazon 2018 cash flow statement shows the non-cash effective capital expenditure in respect of its finance and capital leases. It is good to see this information clearly provided at the foot of the cash flow statement, given the importance for free cash flow analysis.
Amazon supplemental cash flow disclosures given at the foot of the cash flow statement
Calculation of free cash flow by Amazon
Free cash flow is the cash generated by existing business activities after allowing for investment in that business through capital expenditure. It is used as the basis for DCF valuations and for the analysis of the resources that are generated by the business. Free cash flow may then be applied to, say, paying down debt, making dividend distributions or expanding activities through acquisitions.
Amazon states that its “financial focus is on long-term, sustainable growth in free cash flow”. How free cash flow is measured is therefore clearly important. Unfortunately, free cash flow is not defined by accounting standards and therefore Amazon provides it as an alternative cash flow metric. In fact, it provides three different alternative calculations with the difference due to the treatment of capitalised leases. Here is our summary of the three calculations.
Amazon alternative free cash flow calculations for 2018
The first approach is the method most commonly quoted and used in practice: operating cash flow less the net cash flow related to capital expenditure. However, in this approach there is no deduction for any flow related to capitalised leases. In our view this method is deficient in cases where there are significant asset purchases through leasing, as neither the lease payments nor the effective capital expenditure related to new leases are recognised.
The second approach provided by Amazon is the same as above, but after also deducting the capital repayment element of the rental payments for all their capitalised leases (the interest element being already included in operating cash flow).
This method is an improvement and at least ensures that the cash cost of capitalised leases is recognised. Nevertheless, we feel this is still not the best approach as it fails to fully recognise the capital expenditure being made in the period, in whatever way it is financed. It also classifies what should be a financing flow (the repayment of lease obligations) as an investing flow.
In our view, this approach can distort free cash flow where there is high growth, as the impact of that growth on capital expenditure is reflected primarily in future periods and not in the period where the new leased assets are acquired. High growth is, of course, a key feature of the Amazon business. This version of free cash flow includes the financing effect of historical leases, whereas we are more interested in the investing impact of new asset acquisitions arising from today’s new leases.
The third approach provided by Amazon is much better. In this calculation the company excludes the capital element of rental payments in respect of capital leases and instead shows the effective capital expenditure as a component of free cash flow – “property and equipment acquired under capital leases”. In effect Amazon reports the assets acquired under capital leases as though they had been separately purchased and financed with debt.
Here is the full disclosure and explanation they provide for this third approach.
Amazon free cash flow calculation – version 3
This approach is exactly what we advocate and that we explained in our previous article ‘When cash flows should include non-cash flows’. However, unfortunately Amazon only adjusts for its capital leases of equipment and not the capitalised finance leases that largely arise from their ‘build-to-suit’ property transactions. This is in spite of the company providing a disclosure of the effective cash flow for the purchase of assets under both capital and build-to-suit leases at the foot of their cash flow statement. We think it would be better to reflect all effective capital expenditure from capitalised leases in measuring free cash flow. This is one of the three additional adjustments we make below.
The third approach by Amazon is a big improvement on not recognising any effects of leasing in free cash flow at all. It is disappointing that this metric appears to be largely ignored by the data providers. It seems most data providers merely use the standard operating cash flow less capital expenditure (version 1 above) with no additional consideration of the effects of leasing.
Our alternative calculation of Amazon’s free cash flow
Taking the third calculation of Amazon as our starting point we would make three further adjustments to obtain what we consider to be a better measure of free cash flow. Two of these are extending the approach used in version 3 above to all leases. The third adjustment is to apply the same concept of effective flows to stock-based compensation.
Capital expenditure related to finance leases
We asked Amazon why they only make the capital lease and not finance lease ‘effective capital expenditure’ adjustment. Their response was that the build-to-suit leases are not part of ‘core activities’ and that they regard the lease payments themselves as being more representative of cash flow in their ‘normal operating cycle’. This is fine and of course free cash flow can be defined in many different ways. Quite rightly, Amazon stresses in the disclosure we show above that free-cash flow is inevitably incomplete and should only be a component of wider cash flow analysis. However, we disagree and would go further in the free cash flow calculation itself.
We think any asset acquired through lease financing is best regarded as capital expenditure offset by a new financing inflow. We would not differentiate between leases of different types of asset or between leases of different duration. Our first additional adjustment, therefore, is to remove the capital element of finance lease rental payments and replace this with the ‘property and equipment acquired under build-to-suit leases’ shown at the foot of the cash flow statement.
Capital expenditure related to operating leases
In 2018 financial statements, operating leases are not capitalised and the rental expense is included in operating cash flow. The accounting for operating leases will change in 2019 with the majority of these leases then being capitalised. We believe this change better reflects the nature of lease transactions. Many investors already make adjustments to remove the artificial divide between different types of leases.
Our second adjustment is to treat operating leases in exactly the same way that Amazon treats its existing capitalised ‘capital’ leases. Making this adjustment requires us to make some assumptions and approximations. The only information provided under existing accounting standards is the operating lease expense and a schedule of future operating lease payments. By discounting these future payments and then constructing a ‘roll-forward’ of the operating lease liability it is possible to estimate the capital element of those lease payments (our estimate is $2.5bn) and the ‘effective capital expenditure’ arising from new operating leases in the period (we obtain $5.7bn). The higher value of assets acquired compared with lease rentals is consistent with the growth in the business and in operating leasing activity.
It should be noted that these amounts are our approximations. The discounting is inevitably an approximation and, in our roll-forward analysis we have not made any attempt to allow for currency effects or for changes in operating leases due to business combinations.
Leasing is not the only example of transactions that, while not actual cash flows, are in substance two offsetting flows; stock-based compensation is another. The granting of share options or other equity interests to employees involves no cash movement at the time of grant. However, one can regard the transaction as, in effect, the payment of cash to employees with the immediate investment of that cash in the business through purchase of options or shares. Economically a grant of options is the same whether done as a single non-cash transaction or as two cash transactions with offsetting flows.
The reason why we prefer to split share-based compensation into two offsetting flows for analytical purposes is because each leg of the transaction is of a different nature. Cash payments to employees are operating flows and should be included in the free cash flow subtotal, whereas the inflow from issuing equity securities that are, in effect, purchased by employees is a finance inflow and hence not a component of free cash flow.
The amounts involved for Amazon are significant. The ‘non-cash’ add back in the cash flow statement for stock-based compensation is $5.4bn. This is our third adjustment.
Here is our revised free cash flow metric for Amazon.
Amazon alternative free cash flow calculations for 2018
Amazon does a good job in disclosing and explaining their free cash flow measures and we applaud them for recognising the effective flows for some of their leases. However, we believe our alternative calculation of free cash flow better reflects the cash generation by the existing Amazon business. Only if the capital expenditure associated with all new lease financing is included does the true extent of capital expenditure and investment in the business become apparent.
It should be of no great surprise that our free cash flow measure is negative. Amazon is a fast-growing business and significant capital expenditure is necessary to drive that growth. We would be surprised if any capital-intensive company with such a high growth rate were to achieve a positive free cash flow during that growth phase. Investing in a business such as Amazon is surely all about what wealth can be created through investment today and how that can be turned into growing free cash flow in the future as the business matures.
We believe that in order to understand the link between growth, investment, future cash flow and valuation a more comprehensive approach to free cash flow is necessary, particularly the recognition of the ‘effective flows’ arising from leasing, stock-based compensation and other similar transactions.
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 more articles select subject tags:
Or see suggested related articles below: