Sunday 11 February 2018

Expedia and dodgy accounting

I took a quick look at Expedia (EXPE US) last night (yes, this is indeed how I like to spend many of my Saturday evenings). Expedia is one of the world's largest online travel agents (OTAs), and owns a bunch of lodging platform websites you've probably heard of/used - both its namesake expedia.com, as well as hotels.com, Travelocity, Orbitz, Wotif, and HomeAway, as well as air ticket, rental car, and cruise ticket booking operations, and travel media site Trivago. The company's primary competitor is Priceline (owner of booking.com and various other sites), as well as - to a lesser extent - AirBnB and TripAdvisor.

I first looked at the stock last year when it was trading at about US$140, but it immediately looked too expensive for my tastes - particularly with competition on the rise and growth starting to moderate - so I passed and moved on. However, the company reported an earnings miss after-house on Thursday, which sent the stock down (for the second time in 6 months) nearly 20% to US$104. A friend forwarded me a Deutsche Bank research note suggesting I take a look, and two things caught my eye: (1) the stock was down partly because the company guided down 2018 earnings, but to a large extent this was due to a ramp-up in growth investments which would potentially yield additional revenues/market share long term; and (2) the analysts claimed the stock was on a forward EV/EBITDA of just 7-8x.

That seemed too cheap. Travel booking websites are networked businesses and are very profitable and cash generative. 7-8x EBITDA sounded too low. True, competition is heating up (TripAdvisor is moving from pure metasearch into direct hotel bookings, and as the market is starting to mature, Priceline and Expedia have started to compete with each other more vigorously in an attempt to maintain growth rates). However, the company was likely to remain very profitable; the stock was now 30-40% off its highs with deflated expectations for 2018; and according to DB, the company's multiples were now at intriguingly low levels. Was there a potential opportunity here?

It took about 10 minutes reading the company's 4Q17 earnings release and skimming its 10-K to quickly dispel that notion, and uncover the issue: a combination of dodgy accounting, and a shocking lack of critical thinking/scrutiny by the DB analysts. In fact, the valuation is still extremely high (about 40x 2017 earnings), and the stock still looks like a clear avoid to me at current prices, given a slowly maturing market; rising competition; accelerating cost growth; 4Q17 EPS printing just US$0.35 (well below the company's overall 2017 run-rate); and the ever-present threat that deep-pocketed tech giants with large user bases (e.g. Google, Amazon, or Facebook) could one day decide to push into the segment.

There are three fundamental issues with the company's (headline) accounting: (1) the adding-back of stock-based compensation; (2) the adding back of non-goodwill intangibles amortisation; and (3) the capitalisation of software development costs, coupled with the focus on EBITDA. And the company's treatment of some of these items is - frankly - downright dishonest. If these adjustments are not made, the company's earnings multiples approximately double. Specifically, for 2017, 'adjusted EPS' was US$4.30/share, but actual reported GAAP EPS was only US$2.42 - almost 50% less. In addition, for 4Q17, 'adjusted EPS' of US$0.84 (down 28% YoY) was more than twice reported EPS of US$0.35 (down 33% YoY). The first two items above accounted for the majority of the disparity.

I wrote about the reprehensible practice of adding back stock-based compensation here, and Expedia is an archetypal example. The company claims these expenses to be 'non-cash', and added back a US$149m non-cash stock based compensation charge taken during 2017 to adjusted earnings. However, the company spent US$294m in cold hard cash buying back shares during 2017 on market to offset the dilution, and yet still ended the year with a higher share count than at the beginning (138.9m vs. 137.2m). Buying back shares on market with real cash to offset dilution from 'non-cash' stock compensation makes the compensation look very 'cashy' to me. This has not deterred the company from trying to mislead investors about the true level of its profitability, however.

Furthermore, the company added back US$275m in non-goodwill intangibles amortisation during 2017, and yet the balance sheet carrying value of its non-goodwill intangibles declined by less than US$140m. What this means is that the company capitalised more than US$135m of cash expenditure on intangibles during the year (across a variety of categories such as 'customer relationships', 'supplier relationships', 'technology', 'domain names', and 'other'). Some of these intangibles are acquired rather than internally-generated, but they are still finite-life assets that cost real cash to acquire (acquisitions are to some extent nothing more than outsourced R&D). These cash costs never go through the adjusted earnings line - ever - despite the fact that they cost real cash to acquire, and are semi-recurring in nature.

Software development costs are are also capitalised by the company into fixed assets and then depreciated as well. For this reason, 'EBITDA' is actually a fundamentally misleading representation of the company's underlying cash generation ability, because software development costs are a recurring cash expense. The ruse here is to capitalise the cash costs as incurred so as to exclude them from EBITDA, and then to argue that the subsequent depreciation is 'non-cash' in nature, and hence ought to be ignored. That would be less cynical if the company had incurred all of its costs upfront and were no long spending cash on a recurring basis to keep its online properties up-to-date, but that is not the case: software development costs are a major and recurring cash cost for companies like Expedia. Recurring software development cash expenditure is therefore never included in reported EBITDA (let alone adjusted EBITDA) - ever.

I find this kind of dishonest accounting extremely loathsome - perhaps exceeded in intensity only by the disrespect I have for the professional security analysts that do not call the company out on it or include stock-based compensation in their earnings forecasts, valuations, or multiple-based price targets, and run around saying 'Expedia is cheap at 7x EBITDA' (often neglecting to mention that this is 'adjusted' EBITDA). The standards of (some) company reporting and security analysis remain woefully low and need to improve.


LT3000