The Many Different Ways to Guess a Business’s Future Revenue
Geoff's CornerSomeone sent me this email: “Clearly forecasting revenue is very important if we want to forecast earnings/FCF at some point in the future. My question is: How do you think about forecasting revenue for a company? In other words, what do you look at and which are your preferred methods to do it?” It's difficult to predict future revenue. Obviously, the amount of investment the company is doing (just capital accumulation of reinvested earnings in the business) is going to tend to be the biggest driver. So, you can best estimate future sales at a retailer if you know how many square feet, how many stores, etc. they are going to add over time. It is harder to estimate the more important number for investors which is the efficiency of the incremental capital (how many more dollars it drives in sales per square foot or per store and so on). This is unfortunate, because what matters to the investment case is usually how much sales per worker, how much sales per square foot, etc. the company can drive. And yet, if we’re being honest, the easiest way to actually guess at future revenue blindly is to ask the company: how many square feet will you have? How many employees will you have? Knowing how many employees will be working how many square feet of selling space in 203 is exactly that would get you most of the way to knowing what sales will be. That’s the actual truth here. But, it’s not real useful for us as investors. It would only tell us what sales would be. It wouldn’t really tell us what economically valuable sales would be. So, let’s look at some less reliable ways people might actually try to guess at future revenue. There are often industry estimates for likely total market size in future years. You could use these if you assume a constant market share and trust such estimates. I'm not sure if these estimates usually prove accurate or not and can't say I've ever really used them in analyzing an investment. But, if you Google estimates for share of total U.S. electricity generated by wind in 2030, size of podcast market in listeners in 2030, size of corporate A.I. spending by corporations for 2030 – I’m sure you’ll get a hit for every one of those searches. If you want to start there and anchor on that – go ahead. It’s possible that’s a better anchor to start with then your own initially uninformed thinking on the topic. But, it might not be. And it might be hard to get that very speculative anchor out of your head if you do choose to start there. The other logical approaches - again assuming stable market share for the company you’re analyzing (which is a gigantic assumption, but one that’s difficult to avoid at first) - would be things like: Inflation rate Inflation rate plus population growth Nominal GDP growth (basically inflation rate plus population growth plus productivity growth) Something like a supermarket would be expected to be in category 1 (inflation rate) growth in market size over time. People spend less and less on groceries over time relative to what they spend on everything else because groceries are basic (the first thing they buy, the least marginal) and as they get richer they put the additional income into more marginal spending items instead of increasing the basics at the same rate as they increase their income. You might be able to add population growth to this if you know where stores are and expect them to capture the same percentage of the market and there is population growth in those areas. This is a very small difference - and for many companies in this kind of category I would assume no real sales growth over time. So, just revenue inflating at the rate of inflation each year so it grows in nominal but not real terms. Inflation rate plus population growth would be what you'd assume in cases where you expect the real amount of something not to rise or fall over time but the pricing to be stable. For example, you'd expect the market for hair cuts, (basic) restaurants, and other local services to grow at about the rate of inflation plus local population growth. Nominal GDP growth would be what you'd assume with things that tend not to lose share versus GDP over time as an economy develops further, gets richer, etc. Categories like advertising and financial services probably keep pace with nominal GDP. Categories like media and entertainment and so on might do so as well. Law firms, accounting firms, etc. probably do not expect to have a decrease in their proportion of GDP - so they would assume this category 3 type thing. What about things that grow faster than nominal GDP? This is harder to estimate. It means that the product or service is more marginal and bought at higher and higher rates by people as they increase their income. Either that, or - as I'll deal with in a second – something that isn't "fully penetrated". These are growth industries like A.I., cloud computing, streaming (though that is probably going to get close to just nominal GDP growth within the next couple years at most), and so on. For many years, this included spending on education and medical (though since COVID hit, this is no longer true of either of those categories). For a long time, travel grew at faster than nominal GDP and so did eating out (restaurants), though I am not sure that in the most developed markets for these things (like the U.S.) this is really likely to be true in the future. The easiest "growth" industries to analyze would be those that are not fully rolled out on the supply / distribution size rather than those where we expect the share of wallet of each customer to go up. So, say you know people have cable TV in the 1980s and what they spend on it. You know that much of the country does not have cable bypassing their house (the infrastructure hasn't yet been built on their street) so they can't hookup to it. The question then is just calculating how many new homes will be bypassed each year. The assumption is that once it becomes available in the area the economics would be similar to other areas. So, it is just a question of a roll-out and availability. The spending per new customer is expected to be similar to spending of already available customers. You can try to estimate how many people will eventually "adopt" streaming services, podcasts, Amazon Prime memberships, high-speed internet, smartphones, whatever in the roll-out phase of those products where either some people don't know about it that much and lack sufficient familiarity with it and are conservative adopters of new things or situations where the infrastructure needs to be built out. The converse of this is industries that are shrinking in real (though not always nominal) terms. Someone might guess cigarettes for instance will decline an assumed 2% or so in terms of number of smokers each year but increase prices at 3% or more each year. So, real volumes fall (number of units sold is down) but nominal volumes rise (number of dollars collected in sales increases). Very, very gradual societal shifts like this would be fairly easy to project out assuming stuff doesn't feedback through network effects and other things. It's easier to project gradual declines in something like smoking than in printed books, "linear" TV, newspapers, fashion items, health / food items, etc. because these things don't tend to feedback in a group way in a manner that matters much. The more individual and private the demand is and the more driven by individual consumer choices (households are easiest to predict this way) the less likely things will accelerate. Stuff that is a two-sided market (like media) where it depends both on whether audiences want to watch TV a certain way and whether advertisers want to advertise a certain way (this applies to radio, newspapers, etc.) would be more difficult to predict since there are two sides to the market and this can cause feedback in one decline to cause another and lead to cuts by producers in supplying things one way and simply doing it another. Items like fashion and so on where people are driven a lot by other people's decisions would also run this risk. Something people wear outside is different from something people do in private. The more private, more individual, more cut-off from that kind of feedback is the easier to predict. Churn and attrition are major factors to consider as well. Revenue will be easiest to predict where items are subscribed to or purchased extremely frequently at very low cost per transaction etc. versus situations where purchases are infrequent, high ticket price, etc. Habitual use is easier to predict than non-habitual. It's much easier to predict revenues related to fans watching something, people betting on something, people drinking, people smoking, people watching, and so on then it is to predict revenues tied to people going to a gym, dieting, etc. Very few people stay loyally hooked on any diet, exercise, etc. A lot of people stay hooked to what they watch, drink, eat, play, and do in many other ways. Revenue is harder to predict the further away from the core of the business it is. So, the newer and more different it is in geography and product type and so on - the much harder it becomes to predict. For Howden Joinery (a U.K. kitchen cabinets, equipment, etc. supplier) it is easiest to project revenue for existing depots, harder to predict revenue for new depots, and hardest to predict revenue for new stores opened in other countries. We just don't know if the concept will play the same in one country as another. Most products / services will not cross all borders as easily and therefore some locations will not produce as much revenue as home locations did. For chains and expansion like that you usually use the figure for what you think mature stores do, for how long it'll take stores to get mature, and for how many new stores are being opened. This is the most theoretically correct model. But, accurate (not just aspirational figures from the company's investor presentation) figures on this can be hard to get. For that reason, you can approximate this pretty closely by just saying revenue will increase at: Same-store sales growth plus new store growth So, a company grows sales per store at nominal GDP (let's say 5%) in existing locations and it has 100 stores and plans to open 15 more stores this year. For year one, this would be an expected 20% growth rate: 5% same store sales plus 15% (15/100 = 15%). However, this number can't be used out into future years because usually new store growth in percent terms will decline while in terms of number of new stores added it'll stay more constant. So, in this example, it's likely that if the company opens about 15 stores a year every year for a decade, the "new store" percent of revenue growth will start at like 15% a year but decline to about 5% a year over the 10-year period. For that reason, you might want to assume growth in the longer-run is more like 10-20% over periods as long as 10 years (not 20% as it would be in year one). It's easiest to model growth out a few years. However, from a stock perspective - this may not be the most important. There are a few reasons for this. One, revenue growth usually does some "pulling forward" of the eventual potential limits on the company's growth. Wal-Mart, Southwest Airlines, Starbucks, etc. might grow much faster than other businesses in early years. But, some of this is actually that they are just growing much sooner. Eventually, they will saturate their markets and actually grow slower in their huge size years than much smaller banks, insurers, etc. that simply grow 8-12% a year forever and are "growth" companies by the definition of those industries, but which may not exhaust their growth potential for 30-70 years or something (see GEICO which has been a growth stock for most of its existence, as has Progressive, and so on). Stores, restaurants, etc. are easy enough to use a figure for the total number there could maybe be in a given country, region, etc. and use that as an absolute upper limit. This point-to-point calculation is easy to do. You assume the company will reach 300 stores or whatever you think the max size is in 30 years (or whatever you arbitrarily pick as a date they stop expanding). You then just figure out (inflating the number with a likely inflation rate) what that might be in nominal size at that point. Then, you figure out the CAGR needed to get from x in year 1 to however many times x sales will be 30 years from now. This helps be more conservative in your expectations. For stocks, the discounted present value of future cash flows (DCF result) involves a huge amount of value from the out of period years. So, when you choose to cut off the calculation and what you pick as a "terminal rate" matter a lot. Also, the amount of sales growth relative to owner’s capital used to finance that growth is hugely important. For this reason, it would tend to be wildly more important to know a company could grow forever at nominal GDP or higher type rates while using relatively low levels of added capital than it is to figure out things like how many stores it will have built out by such and such a date. Basically, if you could know the product will be as important or more important to society virtually forever, the product will be priced as highly or more highly in real terms forever, and a lot of the growth will be "unreal" (inflation driven), intangible, or financed by someone else (not the company) - this is far, far more important than estimating new store opens for the next few years or something. It's the difference between See's Candies, Moody's, etc. and like Sprouts. You can estimate Sprout’s growth. And the growth does add value. But, there is eventually going to be more of a termination of growth issue with Sprouts than with companies that are not as driven by achieving saturation in physical volumes of product sold. Basically, the growth phase could be shorter and the terminal rate much lower for a company like Sprouts than for a company like Moody's. This would not be obvious at all if you looked out 10 years only (Sprouts might be shown to have higher growth during this period). But, it would have a tremendous influence on the correctly calculated DCF. In practice, investors and analysts doing DCFs don't consider this and just assign pretty arbitrary figures to everything outside the years they are modeling including using unrealistic terminal growth rates for companies. Because of the importance of the extremely long duration of the assets you are buying when you buy stocks and because of the calculation of the price you are getting - I'd say two things are really more important in getting the decision to buy or not right rather than in filling out a model accurately. One: carefully consider near-term revenue (and earnings and so on) growth. Basically, pretend you are doing an LBO of this company or something. What will the next 5 years look like? This is important from a sequence perspective. Basically, you may be able to detect a stock's price is actually cheap versus year 3, 5, etc. earnings compared to other stocks versus their year 3, 5, etc. earnings even though the stock looks like it has a normal P/E today. Two: carefully consider perpetual revenue growth. In other words, what does the revenue growth path look like to a forever owner. This is critically important to divorce from growth that will occur only up to a point. What will growth in the unimaginably far future probably look like. This is key to knowing if the company should always have a high multiple or if the multiple will contract from a growth multiple to a value multiple as the market realizes saturation is nearing. If you get those two things right: 1) what will revenue look like for the next 5 years and what will revenue look like once the company is “done growing", you know the math that does all the heavy lifting in terms of future projections. Being more precise than that isn't very important. It's just important not to get confused and assume that a 15% growth rate for the next 10 years means a high growth rate forever. In many cases, after saturation, a high growth company's growth will be at least as low if not lower than a "no growth" company at that same point in time. But, like I said, equally important is understanding the really near-term growth so you don't get fixated on your "entry multiple". Andrew's CornerYour Weekly Dose of Content & Investing Situations NCR Corp Spinoff
Form 10: https://bit.ly/3rQGLzW VIC Writeup: https://bit.ly/3LZKO3Y INVEST WITH US
|