Why I (Usually) Build Models
Modeling, in the right context, can be a valuable exercise
As you saw in the April 2021 Portfolio Review, I construct five-year financial models for many of the companies that I’m invested in. In terms of the output, the objective of that exercise is to generate a reasonable estimate - albeit a very rough estimate - of per share profitability in the medium term.
The value of this exercise for the long-term investor is two-fold: first, when done properly, it gives us a way to think about today’s valuation relative to the future earnings power of the business; and second, constructing a model draws attention to the important variables in that calculation, which provides insight into which inputs have the most meaningful impact on the output.
Let’s start with the rationale for trying to predict future earnings power, particularly since our estimate will invariably prove inaccurate (sometimes by a wide margin). In a 2016 interview, Tom Gayner, the co-CEO of Markel, said the following when asked about the evolution of his investment philosophy:
“One thing that has changed is a point of emphasis. I’m an accountant by training. When I first started, I was most comfortable looking at financial statements and arriving at a point-in-time assessment of what I thought something was worth today relative to what it was selling for. That’s what card-carrying members of the value investing union do. But what has become more important to me is not the snapshot, but the movie. What do the characteristics of the business say about what happens next? That’s caused me to put primary emphasis on [reinvestment opportunities and capital discipline]. Business is so dynamic today that you’re either getting better or you’re getting worse. I’ve learned through painful experience that without ample reinvestment prospects, you’re generally looking at a rotting asset. Of course, it’s much more fun to buy great companies with bright prospects, so the main reason not to would be because they cost too much. I am still very attuned to that, but at the margin I have become more willing to pay up for the good stuff.”
Another quote, from Bill Nygren of the Oakmark Funds, ties in here:
“For almost all businesses, our crystal ball goes dark after seven years, so we assume all businesses trade at similar P/E's after seven years. With an estimate of fair value seven years in the future, we can discount that back at an appropriate risk-adjusted discount rate… Whether that results in a near infinite or a negative P/E on current earnings is not of concern to us.”
Gayner and Nygren share a similar perception of risk in this circumstance: they don’t want to miss an opportunity to invest in a great business simply because the current P/E is “too high”. It’s a simple idea, one many of us have learned to appreciate (i.e. learned the hard way). But what to do about it in practice, particularly if you have the desire to operate with some valuation discipline? For both, the answer is to adopt a longer-term perspective. For Nygren, that is done explicitly (model seven years out). For Gayner, while he doesn’t share specifics, my guess is that he does so implicitly (accepts a higher current P/E).
Generally speaking, if we can define greatness as the ability to generate higher EPS / per share FCF growth than the average business, this exercise allows us to shrink the “greatness premium”. It focuses our attention on the movie instead of today’s snapshot (and rightly so). That exercise, quantified, requires us to make an estimate of future earnings power (even if the output in some cases will be as vague as “much bigger than today at some point down the road”).
To summarize, a model requires you to define - to quantify - greatness; it provides some perspective on how Mr. Market views a business relative to its normalized earnings power (“his” view on its stability, its growth prospects beyond the forecast horizon, etc.). To use Gayner’s analogy, constructing a financial model is an attempt at taking a snapshot in the future.
The second value-add of modeling is that it helps with thinking through the important variables in that calculation. For example, let’s assume that you’re constructing a model for a retailer like Dollar Tree (a business that will be the focus of Monday’s post). As you model revenue growth for DLTR, your focus in drawn to unit growth and same store sales (an exercise that will need to be completed for both the Dollar Tree banner and the Family Dollar banner). On the first variable, what are reasonable estimates for net unit growth over the next five years – and how much capital / investment will those new stores require? Working through this exercise provides insight on the quality of the business (unit economics), as well as the value (the incremental earnings power) added through unit growth. The same idea applies to same store sales – what’s a reasonable expectation for the next few years (based on past results, competitive dynamics, etc.) and how much investment is likely to be required (for example, into working capital) as the average Dollar Tree increases its annual sales from $1.7 million to $1.9 million (assumes +2% comps p.a.)? This exercise, along with the insights derived from 10 to 20 years of historic financials, is the part of financial modeling that I find most insightful. I think working through this exercise further improves my ability to define and quantify greatness.
As part of this exercise, I’ve also found it’s useful to focus on the quality of growth. Returning to DLTR, what is the breakdown for 2020 - 2025e EPS growth between new units, same store sales growth, net margin expansion, and capital returns (repurchases)? Over time, I’ve learned - painfully - that an overreliance on lower quality sources of EPS growth, like repurchases, is a red flag, or at the very least something to be noted and monitored closely. To paraphrase Gayner, without ample reinvestment prospects, you’re likely looking at a rotting asset.
In summary, I think the outputs from a model should be taken with a grain of salt. As it relates to short-term predictions, the past 18 months provides clear evidence that the quarterly earnings game some analysts - and company’s - choose to continue to play is a fool’s errand. It’s worth remembering that, even you can predict 2021 earnings to the penny, it has no discernable impact on intrinsic value (besides as a signal for the future); the majority of the value from a DCF is dependent upon results that extend well beyond the forecast period.
When modeling leads to a focus on quarterly earnings and pinpoint fair value estimates, it has gone astray. Instead, its the process of thinking about the inputs, and their long-term impact on the outputs, where the exercise can add value for investors. The goal is to ask the right questions on competitive dynamics, addressable markets, and capital allocation - the drivers of long-term value.
Returning to the April 2021 Portfolio Review, the Disney discussion is a good example of how I think about balancing my focus on the outputs of a financial model / DCF with the realities of trying to predict relatively short-term business results with any specificity (both in terms of magnitude and timing). In practice, I think analysts can get overly focused on the reopening date for theme parks and movie theaters, the timing of peak DTC operating losses, etc. While those can be important events, particularly if they have an impact on the short-term viability (financial stability) of the enterprise, they should be of far less concern to the long-term investor than other considerations - for example, thinking through a reasonable estimate for global DTC subs and ARPU’s in 2030.
I think the imperative word when building models is balance.
It’s recognizing what matters to the business and valuation in the short-term, while simultaneously maintaining your focus on a company’s long-term growth opportunities and sustainable competitive advantages. When kept in that context, I believe it can be a valuable exercise for the long-term investor.
NOTE - This is not investment advice. Do your own due diligence. I make no representation, warranty or undertaking, express or implied, as to the accuracy, reliability, completeness, or reasonableness of the information contained in this report. Any assumptions, opinions and estimates expressed in this report constitute my judgment as of the date thereof and is subject to change without notice. Any projections contained in the report are based on a number of assumptions as to market conditions. There is no guarantee that projected outcomes will be achieved. The TSOH Investment Research Service is not acting as your financial, accounting, tax, or other adviser or in any fiduciary capacity.