Modeling Growth Businesses Alexander Motola, CFA Alexander Motola, 2013 1 Biography Portfolio Manager ◦ US Growth, Intl Growth, Hedge Funds, GARP, Growth and Income, Value/Growth ◦ Twice listed in Barron’s “Top 100 Fund Managers”; 7+ Lipper Awards for best risk adjusted performance ◦ 12 year Core Growth record 2.45% annualized outperformance vs. benchmark; 5 year Intl Growth +6.24% Analyst ◦ All Cap Growth, Mid Cap Growth, Small Cap Education ◦ MBA, Haas School of Business (UC Berkeley) ◦ Chartered Financial Analyst ◦ B.A, UC Santa Barbara (English, Medieval Studies, History) Alexander Motola, 2013 2 Modeling Growth Businesses How the markets works; expectations based investing What is “Net Revenue” Overview of Modeling Approaches How does the CEO get paid? (Proxy) What does a revenue model look like? Modeling Tips Specific Examples Alexander Motola, 2013 3 How the market works Efficient Market Hypothesis The weak form (prices on traded assets (e.g., stocks and bonds) already reflect all past publicly available information) is true for most of the market, and is most true for the “best known” stocks The strong form of EMH (all info is instantly priced in) is definitely NOT true; if you thought so, you wouldn’t be in this class Alexander Motola, 2013 4 How the market works The market is expectations based “If we buy the stock today, then each day we move forward, the headlights of the market move forward one day (let's call it). The return that I earn over the next twelve months is the difference between the market's expectations for the first twelve months relative to the expectations it will have twelve months hence.You are looking at an expectation set change one year forward.” – Bill Miller Past data is priced in; how the market thinks about future data is not Alexander Motola, 2013 5 How the market works What does this mean for modeling growth rates of individual companies? If a company grow revenues 20% a year for the next 5 years, and you correctly predicted that, will you outperform the market? Why is modeling growth rates (revenue) important? ◦ Revenue is the lifeblood of growth stocks; it is also the “headliner” for the financial statements Alexander Motola, 2013 6 What is “Net Revenue” According to InvestingAnswers.com, “Net revenue typically refers to a company's revenue net of discounts and returns” Therefore, what you see at the top of the income statement is a number already adjusted by management Every other statement flows from the income statement Only via astute analysis can you determine “discounts and returns”; sometimes it is not even possible to derive this number Alexander Motola, 2013 7 Overview of Modeling Approaches What is “the fade”? Top Down Company Guidance Linear Extrapolation Past Performance Unit Level/Product Level Sequential (QoQ) Cyclicality Deferred Revenue, Waterfalls, etc. The impact of Acquisitions One Special Case: Retailers Alexander Motola, 2013 8 Modeling: Fade (Growth Rates) In theory, all growth rates will become asymptotic to GDP ◦ Rate of fade ◦ Time Matters – is the fade gradual, or is there something which causes a step function (patent expiry, etc.)? ◦ What does the market think, and why? Re-Acceleration is a “holy grail” for investors, because even if the market has the direction correct, it usually is overly conservative on the magnitude Alexander Motola, 2013 9 Modeling: Top Down Relevant GDP growth rates ◦ How fast is the company growing relative to national or global GDP? ◦ Can you use GDP growth rates by country along with revenue mix by geography? Industry Growth Rates ◦ How many players in the industry? ◦ Can you look at all of them? ◦ Read multiple companies’ 10-Ks, etc or industry reports to get industry growth rates Taking or Losing Share? Alexander Motola, 2013 10 Modeling: Company Guidance Companies often provide short or long term forecasts ◦ Earnings Calls, Analyst Days (often webcast); never in their SEC filings Forecasts can be meant for many different constituencies ◦ Competitors ◦ Investors ◦ Other Stakeholders (suppliers, employees, etc) No Accountability, poor accuracy Can be useful as a basis for a high end of range boundry (companies will almost never exceed, but will often fail to achieve their forecasts) Alexander Motola, 2013 11 Modeling: Linear Extrapolation Newton’s First Law of Motion – “An object that is in motion will not change its velocity unless an external force acts upon it” Analysts often do this ◦ It’s easy ◦ It’s “intellectually dishonest” Continues past growth into the future, blindly ◦ Ignores “the fade” ◦ Some projects suceed, some fail Sometimes, in the absence of other data, this can (but not usually) be the “best” approach; however, this can easily lead to an overestimation of future revenues. Alexander Motola, 2013 12 Modeling: Past Performance A close cousin to “linear extrapolation” Uses history as the sole guide Can be useful for understanding the revenue cycles of deeply cyclical industries, but even those usually have some secular growth rate (higher highs, higher lows) If this is used, 15+ years of history should be used, and the margin impact (gross margins) should also be studied (Who covers INTC?) Alexander Motola, 2013 13 Modeling: Unit Level Different business units (“segments”) have differing growth rates Understanding the impact of the growth rates of the various segments can provide a huge advantage in determining the future direction of the total revenue growth rate Companies usually provide a lot of disclosure around segments (and geographies) Alexander Motola, 2013 14 Modeling: Product Level Revenue = Price * Volume (mostly true, there are reserves) Register data, company reported data, Neilsen data, government data – all sources of units sold Sell in ≠Sell through Best opportunity to reach an “out of consensus” perspective on a stock Alexander Motola, 2013 15 Modeling: QoQ Growth Rates QoQ is only useful if you are modeling time periods less than a year (Quarters, Halfs) Linearity refers to how the company collects revenue within a given period (front or rear end loaded) QoQ is very misleading for “seasonal” companies, such as retailers QoQ is very appropriate for highly predictable, recurring style models Alexander Motola, 2013 16 Modeling: Cyclicality Cyclical business experience dramatic changes in price and demand, with huge margin impacts A long history of revenue (20+ years) is useful, along with an understanding of where you might be in the cycle (you must be WELL AHEAD of the cycle to make money) You typically want to buy these when they look the worst “This time” is NOT different 99% of the time Alexander Motola, 2013 17 Modeling: D/R, Waterfalls, etc. Certain business have future revenue on or off balance sheet which can increase the accuracy of any forecasts (keep in mind the delta to expectations drives the stock price) This works best for quarterly forecasting Focus on key drivers to predict business success Alexander Motola, 2013 18 Modeling: Acquisitions Businesses can give a lot of information about acquisitions; you can usually get enough to impute the “organic” growth rate PEP recent 10-K, page 53 has a section called “Organic Revenue Growth” which provides a nice table showing what aspects of their growth are more repeatable than others. Most smaller companies make you do the math or hide acquisitions Alexander Motola, 2013 19 Modeling: Acquisitions (PEP) Will Exchange Rates be the same in the future or move as much as they did in 2012? ForEx helped in 2011. How integral to PEP’s strategy are ongoing acquisitions? Alexander Motola, 2013 20 Modeling: Retailers Certain business/industry models are unique enough to require another method of analysis (retailers, banks, smaller E&P companies) WAG (2Q13 Results Press Release) gives us the following data: Front-end comparable sales, traffic, basket size, total sales. “Pure” Retailers often disclose SSS, comps, new stores, square footage, etc. which will allow a fairly robust model Alexander Motola, 2013 21 How Does the CEO get paid? Proxy Analysis – do bonuses depend on sales growth, ebitda growth, stock price growth, market share? “Reflective of our compensation philosophy, the compensation of our named executive officers is significantly affected by our financial results. As in previous years, the annual non-equity incentive financial performance metric for our named executive officers in 2011 was operating income. Based on the Company’s 2011 operating income of $889 million, which represents a 2.3% increase over operating income in 2010, each of our named executive officers earned slightly above the target payout for annual non-equity incentive awards in 2011. However, in 2011, the long-term equity incentive element of our named executive officers’ compensation was negatively affected by the performance of the Company’s stock price during 2011.” (ALV, 2011 Schedule 14A “Proxy” filing, pgs. 30-31) Alexander Motola, 2013 22 How Does the CEO get paid? What does this mean? ALV goes on to explain the formula: ◦ — Threshold: If the Operating Income is 70% or less of the previous year’s Operating Income, the Company does not pay any annual incentive. ◦ — Maximum: If the Operating Income is 130% or more of the previous year’s Operating Income, the payment equals two times the target amount, the maximum payout under the program. ◦ — Target: Where the relevant Operating Income is between 70% and 130% of the previous year’s Operating Income, the incentive is calculated through linear interpolation (“along a straight line”) between said levels. Incentives matter, so understand what your management is paid to do Alexander Motola, 2013 23 What’s in a Revenue Model? Very simple model of AMG; includes 3 revenue segments Alexander Motola, 2013 24 What’s in a Revenue Model? AMG’s model is pretty basic – AUM * fee for all 3 segments, plus small adjustments for performance fees, so you need to forecast AUM (a function of market performance, marketing, and product performance) and the fee. The last slide had the AUM; here’s the fee history Alexander Motola, 2013 25 Modeling: Tips Differentiate between Fact and Assumption Reduce your key assumptions; simpler is better and often more accurate Forecasting is a flawed “science”; your goal is more to understand what can happen, how it can happen, and “What the market is missing”; you will not forecast an EPS number in the future Track your performance to understand your forecasting errors Alexander Motola, 2013 26 Price * Unit Model: HANS Hansen’s (HANS) is now Monster Beverage (MNST) – one of the biggest misses of my career Relatively unique in that they disclose gross revenues and detailed product level information Typically a company discloses less and less specific information as they get bigger or face slowing growth; they also change definitions, which impacts comparability Alexander Motola, 2013 27 Price * Unit Model: HANS I chose to model HANS based on Case Units and Gross Price per case ◦ Instead of picking numbers, I used growth rates in units and $ (my estimates shown in green, blue italic represents forecasts) ◦ Forecasts supported by other data (see Excel model) Alexander Motola, 2013 28 Organic Growth: MFE (INTC) A lot of investors owned MFE because mgmt claimed the organic growth was fairly high; I didn’t agree at high prices because I thought the organic (non-acquisition) growth was much lower. In fact, my best guess is it was 0% Revenue Model (R166- 357) R193-243 focuses on acquisition analysis; it shows revenue contribution from various acquired companies or combinations of acquired companies For example, MFE bought SCUR+Reconnex+Solidcore (R208-209); look at 3Q09, if they had those companies in 2Q09, revenue growth would have been 3.6% yoy, instead of the reported +18% (R325) Alexander Motola, 2013 29 Mix Shift: ZOLL Exciting new product is growing much faster than corporate average, becoming a bigger and bigger part of revenue each quarter, and moving the corporate total growth rate higher (notice how fast LifeVest grew each Q) Zoll reported 1Q09 on 1/22/09; stock was $16.54; by 7/28/11 (3Q11 report) stock was at $69.66; if you caught this mix shift, you made a lot of money; stock was finally acquired by Asahi Kasei on 4/23/12 for $93 as LifeVest continued to grow rapidly Alexander Motola, 2013 30 Retail Model: CAKE Alexander Motola, 2013 31 Waterfall Model: RNOW Excel Model R8-23; 2Q11 beat and guidance R206-222; key metrics R280+ Revenue model ◦ ◦ ◦ ◦ ◦ Segment & geography (R281-318) OBS backlog tracking (R320-328) Waterfall (R347-413) Deal Metrics, Beat History, D/R Start of tracking forex impact (R494-520) Alexander Motola, 2013 32 Summary What’s priced in? (expectations) Modeling helps you understand what makes the business work Different Techniques for Different Business Models Keep things simple, estimate the fewest variables in your forecast Fit your projections to the data, not the other way around Focus on what’s material Alexander Motola, 2013 33