At Morningstar, we assign fair value estimates to around 1,800 companies across the globe. Each of these fair value estimates is based on a rigorous discounted cash flow (DCF) model built by one of our analysts using a standard Morningstar template. Occasionally, we find ways to improve our methodology. Over the next several months, we will be rolling out the sixth generation of our internal DCF template. In this article, we describe some of the key features of our updated valuation methodology.
Changes to our methodology may require adjustments to some of our fair value estimates, which you may notice in the coming months as analysts transition their companies to the new model. Some of these changes will tend to increase our fair value estimates, while others will cause our fair value estimates to decline.
Morningstar's Three-Stage Discounted Cash Flow Valuation
Our DCF model includes three stages of analysis. The first stage includes our forecasts for the next five to 10 years. In the first stage, analysts make explicit forecasts for all of a company's important financial statement items, such as revenue, operating costs, capital expenditures, and investments in working capital.
In the second stage, analysts are more limited. We take earnings from the last year of Stage I and assume that they grow at a constant rate. We determine the investment needed to achieve this growth by assuming a constant return on new investment. Analysts are responsible for choosing the growth rate, the rate of return on new investment, and the length of Stage II, but otherwise don't need to make explicit forecasts for individual financial statement lines.
Stage II assumptions are the primary vehicle for incorporating our analysis of economic moats in our fair value estimates. Companies with wide and narrow moats are expected to earn returns on new invested capital that exceed their cost of capital in Stage II. The wider the moat, the longer Stage II is likely to last. In general, we assume narrow-moat companies can earn excess returns on capital for at least 15 years, while wide-moat companies can earn excess returns on capital for at least 20 years.
Our model concludes with a third stage. In Stage III, all companies are assumed to be the same. Return on new invested capital is set equal to the weighted average cost of capital; every moat is eventually eroded--no company can earn excess returns forever. We also assume returns on existing invested capital remain constant in Stage III.
Our latest model includes several alternative Stage II-III methodologies, as well. These include terminal multiples (such as EV/Sales and EV/EBITDA) and the ability to enter the total value of cash flows beyond Stage I directly. These alternative approaches should only be used in special circumstances where the standard three-stage method would be inappropriate.
A change to our formulas for valuing Stage II and Stage III cash flows will have the largest downward effect on our fair value estimates, particularly for companies where a significant portion of value is concentrated in these later periods. This is because of more conservative assumptions for long-run reinvestment needs relative to previous versions of our model.
Estimating the Cost of Capital
We discount future cash flows using the weighted average cost of capital, which incorporates the cost of debt, equity, and preferred capital. The discount rate is a key assumption in any DCF model. While the cost of debt and preferred stock can be observed in the marketplace, the cost of equity presents a significant challenge. In the past, analysts have been allowed significant discretion in choosing a cost of equity (COE), but we have formalized our approach in the latest model.
The most common methodology for estimating the COE in practice is the Capital Asset Pricing Model (CAPM). However, we find that the CAPM raises more questions than it answers by replacing one unobservable input (the cost of equity) with three (the risk-free rate, the equity risk premium, and beta). Even among experts, there is significant disagreement about appropriate values for the equity risk premium and beta.
Since we believe our analytical advantage is in estimating cash flows rather than making precise estimates of inherently unknowable quantities, we have chosen a greatly simplified approach that still captures the essence of the CAPM. We will be assigning each of the companies in our coverage universe to one of four "systematic risk buckets." For companies based in the U.S. and several other developed markets, below-average systematic risk will correspond to an 8% COE, average to a 10% COE, above average to a 12% COE, and very high to a 14% COE. Some international markets will require that a premium be added to these values, currently ranging from -1% for Japan to +11% for Greece.
Holding all else equal, we expect our enhanced cost of capital methodology to result in modest increases to most fair value estimates. However, in some cases where companies are deemed to have above-average systematic risk, it is possible that the new methodology could result in slight downward pressure on some fair value estimates.
Accounting for the Time Value of Money
The final significant change to our methodology involves the time value of money. Discounted cash flow valuation produces an estimate of a company's worth as of a specific point in time. That value tends to increase over time as cash flows are earned and future cash flows are discounted less.
In the past, fair value estimates in our models have adjusted continuously, with the published fair value estimate representing the valuation as of the day of publication. Unfortunately, this means that our fair value estimates can become stale as time elapses between report updates. It can also make it difficult for analysts to parse the causes of a change in a fair value between altered assumptions and the time value of money.
We are enhancing our time value of money methodology so that in the future, our fair value estimates will refer to the end of the current fiscal year. Fair value estimates will be updated for time value of money only once per year, when we roll our models. This should make our fair value estimates more forward-looking as well as provide better clarity around the causes of fair value changes. In isolation, this change would tend to increase our fair value estimates modestly.
