A couple of days ago, Paulo Santos wrote an article where he argued that Amazon's (AMZN) cash flows were artificially high as operating leases have increased over the last couple of years. Rather than investing in capital expenditures (which decreases the free cash flow), Amazon leases some of its "assets" and the expenses shows up over time rather than in one year.
Source: Paulo Santos
Santos argued that had Amazon capitalized the leasing commitments instead, the net effect would be this:
Operating margins would be higher, net margins would be about the same (because you'd have amortization / depreciation costs and interest replacing the operating lease). Operating cash flow margins would seem quite a bit higher as well (because depreciation gets added back into those)
While this statement works as a general rule, it can differ from year to year. To be more specific, the effect that leasing commitments has on the earnings and cash flows of Amazon depends on the following factors:
- Depreciation method
- Interest rate
- The duration of the leasing commitments.
Since Amazon's margins are very small, it is in my opinion very important that we try to be as precise as possible when we estimate the effect of the leasing commitments.
Therefore I have looked into the SEC filings, and I found the following statement:
We lease our corporate headquarters in Seattle, Washington. Additionally, we lease corporate office, fulfillment and warehouse operations, data center, customer service, and other facilities, principally in North America, Europe, and Asia
Now that I have a decent understanding of what kind of assets they are leasing, I need to know the depreciation method that Amazon would have used, if they had capitalized the leasing commitments. In the same SEC filing, I found the following statement:
Fixed assets include assets such as furniture and fixtures, heavy equipment, servers and networking equipment, internal-use software and website development. Depreciation is recorded on a straight-line basis over the estimated useful lives of the assets (generally two years for assets such as internal-use software, three years for our servers and networking equipment, five years for furniture and fixtures, and ten years for heavy equipment).
Given that most of the leasing comes from fulfillment and warehouse operations, I think that it is fair to assume that the average lifetime of the leases are roughly seven years.
The next thing I need to estimate is the interest rate. Below you can see the implicit interest rate Amazon uses on their capital leases, which I have calculated to be (49/1,024) 4.7%.
In the below table you can see the operating lease commitments for the next five years and "thereafter."
Given my knowledge of the leasing commitments, interest rate, and depreciation method, I can calculate the present value of the future leasing commitments and quantify how a conversion to a capital lease would affect the income statement. As you can see in the below table, the costs related to the potential "capital lease" was $282.6 million in 2011
The present value of the future lease commitments increases for each year. In 2009 it was $947 million, 2010 $1,482 millions and in 2011 it was $2,538 million. The capital expenditures (CAPEX) can be calculated by taking the difference between the sum of PV commitments of the current and the prior year and adding the lease commitment for the specific year. As you can see below, CAPEX would have been much higher if the operational lease was reported as a capital lease.
The Income Statement
Another thing we can do is to adjust the income statement by removing the reported operating lease costs and instead subtract expected depreciations and interest rate from net earnings. As you can see in the below table, net earnings for 2011 would have been $736 million instead of $631 million, and the operating margin would have been higher as well.
Effect on the Free Cash Flow
So how do these changes affect the free cash flow? Given that the "true" CAPEX figures are heavily understated, it has a negative impact. In the below table, you can see that the adjusted free cash flow is much lower than the reported free cash flow. Therefore we can conclude that the TTM P/FCF ratio of 106.5 is heavily inflated, but the "true" P/E-ratio is also lower than the reported ratio of 312.5.
Effect over the last 12 months
Unfortunately, the future leasing commitments are only reported on an annual basis, which means that I have to estimate the effect on the results over the last 12 months. In the below graph, you can see my estimate, which is based on the trend over the last years.
Since reported TTM FCF was $1,099 million, the adjusted FCF has been negative over the last twelve months. Adjusted net earnings, however, would be $611 million, which is equal to an EPS of 1.32 and a P/E-ratio of 191. Still extremely high, but less "insane," than if we didn't adjust for operating leases.
So how does this affect your investment? Well, assume you are an investor who primarily looks at the P/E ratio. After having done your due diligence on Amazon, you have estimated that Amazon is currently overvalued given its expected future growth rate, and you believe that a fair P/E-ratio is 150. But if you adjust for operating lease commitments, "true earnings" are actually slightly higher, and therefore the implicit fair value of your P/E estimate is $203 for one share of Amazon. Had you not adjusted, however, the fair value of one share would have been $150.
If cash flows are more important to you than earnings, then adjusting for leasing commitments makes Amazon look like a terrible investment, as adjusted free cash flows are negative. So Paulo Santos definitely has a point that Amazon's cash flows aren't what they appear to be.