Model Portfolio Update: What is Value in Tech?

sagawa
Print Friendly
Share on LinkedIn0Tweet about this on Twitter0Share on Facebook0

SEE LAST PAGE OF THIS REPORT Paul Sagawa / Tejas Raut Dessai

FOR IMPORTANT DISCLOSURES 203.901.1633/.901.1634

psagawa@ / trdessai@ssrllc.com

twitter.jpg @PaulSagawaSSR

March 4, 2019

Model Portfolio Update: What is Value in Tech?

Three months ago, after a hard market correction which saw the tech components of the S&P500 drop nearly 16% from their peak, we predicted that growth tech would lead a strong recovery. So far, so good. Our model portfolio is up 19.9% since December 4, beating the S&P by 1608 bp and the tech components of the index by 1300 bp – one of our best quarters since its launch 7 years ago. Of course, this performance leaves many investors worried for valuation, but we are not so concerned. In DCF, perhaps the most widely used methodology, 70% or more of a company’s value is in its year 10 terminal value, itself driven by unquestioned assumptions about growth and profitability beyond the explicit cash flow projections. Stocks that were widely questioned as expensive in past years have dramatically outperformed even their most aggressive estimates and should have merited much more generous assessments of terminal value years ago. Given this, we are less concerned with traditional value metrics, historically poor predictors of tech stock performance, when evaluating growthy TMT names. Rather, we believe the potential for a company to exceed the narrative embedded in its consensus estimates and share price is the most important driver of performance in the sector. In that context, we are removing NVDA from our model portfolio in favor of PVTL. While we still believe strongly in NVDA’s long term upside, we are concerned that its 2019 results may not reflect its true potential and distract from the narrative.

  • Our model portfolio has dramatically outperformed. Our 15-stock model portfolio appreciated 19.9% since our Dec. 4 update, outpacing the S&P500 by 1608 bp and the tech components of that index by 1300 bp. This is the best performance against the tech benchmark in the 7 years that we have published the portfolio, reflecting a very strong bounce back from the 4Q18 market correction led by the growth tech names that predominate in our choices. Performance has also been strong over the longer term. We beat the S&P500 by 3617 bp over 12 months, 9299 bp over 24 and 12592 bp over 5 years, and the tech elements by 3307 bp, 6145 bp and 6245 bp respectively.
  • Strong performance despite “expensive” stocks. The average P/E of our model portfolio has been at least 2.5 times the market P/E over the past 5 years. The P/S has been 3.7 times the S&P500, while the average PEG ratio has been 1.7, well north of the 1.2 of the broader market. At the same time, our portfolio has seen upside surprises out pace misses 7.5 to 1 on EPS and 3 to 1 on sales. This jibes with our experience running a quant model as a small cap strategist – value metrics are almost universally poor performance predictors for high growth stocks.
  • DCF models are particularly suspect. Theoretically, the price of an asset ought to be the value of all future cash flows from the investment, discounted at the cost of capital back to today’s currency. In practice, analysts do explicit cash flow forecasts for a period– generally 5-10 years – and then assign a terminal value, based on final year cash flows and perpetual growth at the pace of the overall economy, or on an assumed market multiple. Typically, 70% of the value of a stock is in this terminal value, so the assumptions used to calculate it are extremely important. Unfortunately, those important assumptions are usually arbitrary and are often manipulated to suit an analyst’s narrative.
  • Successful tech companies beat even the most aggressive projections. The range of analyst estimates for growth tech are almost always far too narrow. Today’s leaders – AMZN, AAPL, MSFT, GOOGL, etc. – have sported lofty multiples of trailing sales and earnings for years, but dramatically out earned projections over the past 5-, 10- and 15-year intervals. With 20/20 foresight, all should have traded at significantly higher market cap, presuming much higher terminal values than were likely being assigned. Of course, there is a strong survivorship bias – once favored companies have also failed – but these circumstances clearly reward the investors best able to separate winners and losers.
  • Picking winners starts with understanding technology. The modern history of technology is one of distinct eras and generational paradigm change. The 60’s and 70’s were dominated by IBM mainframes, Ma Bell, and traditional media. The ‘80’s brought the PC revolution, the first cell phones, the AT&T break-up and the rise of Cable TV, bringing MSFT, INTC, CSCO, DELL, NOK, CMCSC, ORCL, and many others to the lead. The new millennium brought the bursting of the internet bubble and the rise of smartphones, with cloud datacenters, ubiquitous wireless networking, and AI software ascendant. Understanding the opportunities created by these new paradigms and the companies best positioned to exploit them are the keys to finding stocks that will exceed estimates.
  • Our model portfolio stocks are positioned to benefit from thematic change. We have written in depth on SaaS applications (SaaS: Handicapping the Unicorn Races), hyperscale datacenters (Hyperscale semiconductors: Processor diversity coming to cloud), streaming media (The End of TV: Media strains to adapt to streaming era), 5G wireless (5G: Rising global carrier competition to drive capex5G: Hyperscale, AI, 5G sea change), Blockchain, and AI in many forms (AI, Blockchain and The Cloud: Banks in the new technology eraAI Assistants: The new ui paradigm that will end the OS-App EraAI Risk Management: Catalyzing bankings move to the cloudSelf Driving Cars: Building a team to bring TaaS to market). Our portfolio constituents reflect our best picks to play these formative technologies.
  • Sometimes, companies make mistakes. While we are focused on how each company is likely to fare in the context of broad generational change, execution also matters, particularly in setting the near-term narrative around the stock. Tech stocks that mess up typically find themselves in the penalty box until they can reestablish their success trajectory. In this context, we are removing NVDA from the portfolio, not because we are less confident in their long-term dominance, but because we believe that the self-inflicted channel inventory issues and difficult YoY compares will keep the shares trading sideways for a few more quarters. Similarly, we had planned to add NTNX in its place, but that company’s recent report shows a very clumsy transition as it tries to shift to a subscription revenue model. We are concerned that this will take a couple more quarters to sort out.
  • Adding PVTL. Instead, we are adding PVTL, which offers development software for building cloud-native software for hybrid enterprise environment. We recently wrote about the emerging opportunity for hybrid-cloud, multi-cloud platform software (Hybrid Cloud Platforms: Herding cats and dogs together efficiently and securely)and see PVTL as a key beneficiary and potential M&A target.

Our Model Portfolio Performance

For the roughly 3 months since our last update on December 4, 2018, our 15-stock model portfolio appreciated by 1992 bp, beating the S&P500 by 1608 bp and the tech components of that index by 1300. This is the best quarterly performance vs. the tech index that we have posted in the 7 years that we have published the portfolio (Exhibit 1). This easily recouped the 4.6% decline seen in the August to December period (N.B. that decline still beat the S&P500 by 34 bp and the tech index by 467 bp).

Leading the way were 5G related stocks KEYS (+40.9%), XLNX (+39.2%), CIEN (+36%) and SaaS stocks ZEN (+37.3%), NOW (+34.7%) and AYX (+26.3%), with NFLX (+29.8%) also a notable outperformer. Meanwhile, ACN (-0.1%) and NVDA (-0.3%) were all down on the quarter, with MSFT and AMZN also underperforming the broader market (Exhibit 2). Of these, NVDA which preannounced disappointment for its 4QFY19 results is the most concerning, given its ongoing work to reduce a glut of inventory in its distribution channel juxtaposed with extraordinarily difficult YoY compares.

Looking further back, our model portfolio has outperformed over longer intervals as well. Versus the S&P500, it is up 3617 bp over 12 months, 9299 bp over 24 months and 12592 bp over 5 years. Against the tech components of the S&P, it is up 3307 bp, 6145 bp and 6245 bp for 1, 2 and 5 years respectively (Exhibit 3, 4). We also note that we restrict portfolio changes to published updates (generally once per quarter) and that the constituents are rebalanced to equal weighting with each update. The intent is to provide investors with a palette of recommendations in support of our thematic research.

Exh 1: TMT Model Portfolio performance relative to benchmarks since last update

Exh 2: SSR TMT Model Portfolio Constituents – before update

Exh 3: TMT Model Portfolio cumulative performance relative to benchmark – 5 Yr

Exh 4: TMT Model Portfolio nominal performance relative to benchmark – 5 Yr

You Get What You Pay For

Growth stocks trading at lurid multiples of sales and non-existent earnings draw heavy skepticism from investors trained on traditional valuation techniques. While we no longer have access to the sophisticated quant models that we used in our stint as a small cap strategist at a major brokerage, that experience added analytic reinforcement to our conviction that value metrics were of little use in predicting tech winners.

Our model portfolio is and has been filled with “expensive” stocks that have been the primary drivers of its outperformance. The average trailing P/E of our 15 stocks has been more than twice that of the broader market in all but two quarters of its existence (Exhibit 5). Similarly, price to trailing sales has also been 3.7 times the S&P500 on average over time (Exhibit 6, 7). The P/E to growth ratio (PEG), also based on trailing earnings, has been above the market in all but two quarters in 2016. However, these “expensive” stocks have also delivered average quarterly upside sales and EPS surprises 100% of the time over the last 5 years. (Exhibit 8, 9). This is the factor that has driven outperformance.

DCF Will Say Whatever You Want it to Say

Economic theory holds that the intrinsic value of a financial asset is the total of its future cash flows discounted back to present day currencies according the investor’s cost of capital. The modern discounted

Exh 5: TTM P/E Ratio trend for TMT Model Portfolio vs. benchmarks over 5 years

Exh 6: PEG Ratio trend for TMT Model Portfolio vs. benchmarks over 5 years

Exh 7: TTM P/S Ratio trend for TMT Model Portfolio vs. benchmarks over 5 years

Exh 8: Surprises posted by model portfolio constituents in most recent quarter

Exh 9: Historical quarterly surprises by model portfolio constituents last 5 years

cash flow (DCF) methodology was first laid out in Irving Fisher’s 1930 book “The Theory of Interest” but the concept had been in practice for centuries. It is fair to say that DCF valuation has stood the test of time, and for investments that have clear, predictable and finite patterns of future payoffs, it is the obvious tool of choice. However, for investments with wide ranges of potential outcomes, DCF models can be very misleading.

In practice, Wall Street analysts typically predict 5 to 10 years of explicit cash flows for a company, then rely on a formulaic terminal value – usually growing the final year cash flows at average economic growth for perpetuity or applying a market average multiple, then discounting it back to present dollars – to come to a calculation of “value”. For a stock that is predicted to grow its cash flows by 10% per year for 10 years, this methodology still derives more than 70% of the total value from that terminal value. Even if the explicit cash flow forecasts are accurate, and they rarely are, the leverage of the terminal value provides acres of room to manipulate the DCF calculations (Exhibit 10, 11, 12, 13).

Moreover, for companies that take leadership in generational change, a 10th year terminal value that presumes final year cash flows that stagnate at the growth of the economy or that are worth an average market multiple may be inadequate. Looking at top stocks, like Amazon, Microsoft, Apple, or Google, market caps from 5, 10, or 15 years ago imply miserly terminal values when considered with the actual cash flows generated by those businesses. These stocks were wildly undervalued in years past – even the most aggressive analysts missed their actual returns by 20% or more – highlighting the risks inherent in applying a technique that rests on accurate foresight so far into the future. Of course, these examples are the winners –

Exh 10: AMZN Implied and Actual value using DCF and perpetuity growth rate

Exh 11: MSFT Implied and Actual value using DCF and perpetuity growth rate

Exh 12: AAPL Implied and Actual value using DCF and perpetuity growth rate

Exh 13: GOOGL Implied and Actual value using DCF and perpetuity growth rate

survivorship bias hides the biggest failures (the worst of which have ceased to exist) from consideration – but this is another cause to question the usefulness of DCF for growth tech.

Beat and Raise

While value metrics do not work for tech stocks in quant models, upside surprises and upward revisions are powerful drivers of performance, even after the fact. If an investor can find stocks in the sector more likely than not to beat expectations, that is a winning strategy.

We believe finding winners in TMT requires understanding the implications of technological change (Exhibit 14). What innovations are disrupting existing paradigms? What are the business implications of those disruptive changes? What companies are best positioned to benefit from those implications? Our thematic research supports our thesis of generational change driven by the hyperscale datacenter architecture at the core of cloud-based computing, by the machine learning programing techniques behind the rise of artificial intelligence, by the ubiquity of smartphones and the extraordinary wireless connectivity enabled by 5G. These tectonic paradigm shifts have profound implications for businesses across the economy – eCommerce, streaming media, autonomous vehicles, digital payments, dramatic advances in healthcare, and many others. We have tried to build our portfolio from companies best positioned to exploit these opportunities (Exhibit 15).

Exh 14: SSR’s Investment Themes focus on generational changes in TMT