Commentary

Key Takeaways

• The November election led to changes in our portfolio, more because of how the market reacted to it than because we updated our assumptions about tax rates, health care policy, or GDP growth potential.
• Valuation is the core of our investment process, and discounted cash flow (DCF) modeling is our primary tool for understanding valuation.
• We don’t simply create one scenario that we think will happen, however. We construct our models to be flexible in the case of changes in important line items.

Market Review & Outlook

Economic conditions in the U.S. are benign, but stock selection is becoming increasingly challenging, as valuations generally embed expectations for significant value creation. After a very long bull market, we have put away the shotgun in favor of the sniper rifle. Fortunately, small cap stocks offer a rich variety of targets hidden in out-of-the-way places. We continue to find investment opportunities in a wide range of industries, including pharmaceuticals & biotechnology, where we have long been underweight.

The November election led to changes in our portfolio, more because of how the market reacted to it than because we updated our assumptions about tax rates, health care policy, or GDP growth potential. While the likelihood of changes to tax and health care policy has risen, we think it is too soon to make major changes to the odds of specific outcomes, because the details of any plans have been limited and the political environment remains catatonic. The market, however, quickly made meaningful changes to its assumptions. During the period from Nov. 8, 2016, to March 31 this year, bank stocks in the Russell 2000 Index soared as much as 31%, materials as much as 24%, and oilfield services as much as 37%. Those moves in generally commodity industries prompted us to reduce exposure to those industries, moving further underweight in materials and banks and taking an overweight to an underweight in energy.

We redeployed the capital primarily into two sectors we had previously been underweight, health care and real estate, though the stocks chosen were not necessarily reflective of those sectors. In health care, we added a specialty generic drug maker, Akorn, as well as HealthSouth, a provider of rehabilitation centers, and two commercial-stage drug companies, Lexicon Pharmaceuticals and Keryx Biopharmceuticals. In real estate, we bought Realogy Holdings, a brokerage service provider. Our cash balance also grew, as we look for further opportunities to deploy capital.

Valuation is the core of our investment process, and discounted cash flow (DCF) modeling is our primary tool for understanding valuation. We are often asked why we use DCFs, with their many inputs and reliance on important variables that are somewhat subjective, such as the cost of capital. Our answer is: It’s the best tool we have for contextualizing valuation. The DCF can’t simply tell you the answer, but it can help you understand better how the company’s financial model works, what variables are the most critical, and what assumptions go into any given value for the stock.

DCFs have a checkered reputation. While most financial analysts agree that a company’s value can be described as the present value of its future cash flows, many people consider DCFs overly precise. After all, how can we accurately forecast, for example, the gross margin for a company seven years from now? Proponents of DCFs often add to this perception by creating elaborate, incredibly complex models that may magnify the potential for error, yet simultaneously give users greater confidence in their forecasting ability. Business school professors don’t help this perception, as they drill into exercises like capitalizing leases in the future, which adds little value in a forecasting model. As a result, many practitioners think it makes more sense just to "slap a multiple" on sales or operating/net income a few years out and call it a day.

"Don’t use DCFs, to paraphrase Scottish poet Andrew Lang, as a drunk uses a lamppost - for support rather than illumination."

This reaction misses the point of DCF modeling. First, sales or earnings multiples carry with them all the forecasts that DCFs do - what future growth and profitability will be - but without the transparency. Imagine constructing a complicated model that yields a value of 15x 2017 earnings per share. That’s the model that equals the multiple you "slapped" on. All of those assumptions are packed into the multiple, but you don’t know what they are. Second, one should not look to a DCF to give a definitive answer. Everyone who has created a DCF knows that they can say anything you want them to - garbage in, garbage out. Don’t use DCFs, to paraphrase Scottish poet Andrew Lang, as a drunk uses a lamppost - for support rather than illumination.

DCF modeling is a critical tool in our investment process. It is important that the model be economically sound, but beyond that, as Einstein said, “Everything should be made as simple as possible, but no simpler.” Through DCFs we determine approximately what the market expectations are for a stock (i.e., what level of reinvestment, incremental return above the cost of capital on that investment, and for how many years, would make the stock worth its current price). We also use DCFs to form our own view of the company’s value and identify specifically what factors (sales growth, margin, investment rate, etc.) are more likely to be better, in our view, than the market’s expectations. In this way, we can contextualize the company’s valuation robustly in terms of future financial performance.

We don’t simply create one scenario that we think will happen, however. We construct our models to be flexible in case of changes in important line items. Then we create multiple scenarios for valuations that reflect different future business conditions. For example, we may use a decision tree to assign odds of a rapid rise in interest rates and what the valuation of the company would be if that happens. Then we assign odds and valuations to other scenarios - a slow rise in rates, flat rates, falling rates, etc. - and figure out what the probability-weighted value is for the company. We also may use a Monte-Carlo simulation in our models to create a distribution of possible values. The shape of the distribution can take many forms, the most familiar of which is something like a bell curve. Decision trees and probability-weighted distributions of value can look very quantitative and machine-made, but they are the result of fundamental analysis of the company’s competitive strategy and value drivers. At this point, we can discuss as a team both the major assumptions behind different valuations and the odds of those outcomes, allowing for greater cognitive diversity in our analysis.

"By making the conversation about odds more than the analyst’s point of view, there is less reason for the analyst to be defensive about their position."

This approach carries other advantages as well. By making the conversation about odds more than the analyst’s point of view, there is less reason for the analyst to be defensive about their position. It also can reduce anchoring, since the range of values incorporates the uncertainty the future holds, so the analyst can see how the change in facts may mean a different outcome. In addition, the shape of the distribution informs us about the risk of the investment, which helps for position sizing. Overlapping the distributions in the portfolio, together with the correlation of the stocks, can help us understand how we may be compounding a certain type of bet, even when the companies reside in different sectors.

DCF models are analytical tools, but they can function as a check on behavioral errors. We try to use them to reduce overconfidence, anchoring, and endowment effects. It makes updating our beliefs as we get new information much easier. The model itself will not answer definitively what the stock is worth, but it can lead to far better understanding of the assumptions behind a position.

Portfolio Highlights

The ClearBridge Small Cap Strategy had a positive absolute return for the first quarter, outperforming the benchmark Russell 2000 Index. Relative portfolio outperformance was driven by security selection effects and the interaction of sector allocation and security selection. Sector allocation effects detracted from relative performance. In particular, stock selection in consumer discretionary, energy and consumer staples contributed to relative returns in the first quarter. Meanwhile, stock selection in the health care, materials & processing and technology sectors hurt relative performance.

In terms of individual securities, Triton International, Gray Television, 2U, Sprouts Farmers Markets and Q2 Holdings were the largest contributors to performance, while the biggest detractors included ICF International, Horizon Global, CSG Systems, Bankrate and Clearwater Paper.

Over the course of the quarter, we initiated 10 new positions: Aaron’s, Akorn, Foundation Building Materials, HealthSouth, Horizon Global, Keryx Biopharmaceuticals, Lexicon Pharmaceuticals, Realogy Holdings, REV Group, and RingCentral. We also exited eight positions: Radius Health, Steel Dynamics, Criteo, Copart, Surgical Care Affiliates, Hi-Crush Partners, Red Robin Gourmet Burgers, and Taylor Morrison Home.