Projecting Revenue Growth (Historical)

Quick Summary
1. Calculate the historical geometrical average growth rate of the company’s revenue.
2. Run a log-linear regression model of revenue over time.
3. If expected inflation is significantly different from inflation in the recent past, calculate the historical real rate of growth and then add projected inflation.
4. Compare company’s historical growth rate to competitors and the industry average.
5. Determine the length of time for the short-term growth rate.
6. Pick a terminal growth rate.

1. Historical Geometrical Average Growth
Forecasting a company’s revenues is arguably the most critical assumption in a valuation context. It is also the most difficult assumption to make. When divining the future, an analyst must decide among different approaches to modeling revenue. These methods generally fall into one of two categories: top-down or bottom-up (or some hybrid of the two).

A good place to start when projecting a company’s revenue growth is to review its past growth (typically 10 years). This is an example of a bottom-up approach. Since the arithmetic average weights percentage changes in each period equally it can give a false impression of past growth. This is especially true when year-to-year growth has been erratic. A more accurate measure of growth in the past is given by the historical geometric average, which additionally takes the effect of compounding into account. For instance, the geometric mean of 7%, 3% and 2% is [(1+7%)x(1+3%)x(1+2%)]^(1/3)-1=3.98%. (For graphical representations, three-year rolling averages can also be used to moderate fluctuations).

2. Regression Model
Although the geometric mean considers compounding, it ignores any trends that may have developed in growth rates over the period. This can partially be overcome by using regression models. The dependent variable (known y’s) is the natural logarithm of revenue (specified in dollar terms) and the independent variable (known x’s) is the time period (e.g. year). The slope coefficient of the log-linear regression becomes the measure of growth.

3. Real Rates of Growth
A possible modification to the above calculations is to analyze growth results using real rather than nominal revenue to strip out the effect of inflation. This involves dividing each year’s revenue figure by the Consumer Price Index (CPI) of the same year and then calculating year-over-year growth. You must then add expected inflation to the real results to derive growth projections. (There is an example of how to calculate expected inflation elsewhere on this blog, here). A history of the CPI can be found on the Bureau of Labor Statistics website (here). Open the most recent Detailed Report and search for Table 24.

4. Review Competitors and Industry Averages
Of course past growth is not a reliable indicator of future growth. Revenue growth is at the very least more persistent and predictable than earnings growth. Special consideration should be given to how the future growth rate will deviate from the recent past. One of the greatest challenges facing an analyst is anticipating inflection points where the historical trend starts to bend. Be sure to compare the company’s past to the historical results of its competitors as well as carefully consider what systems the company has in place to scale successfully. It is additionally worth paying attention to the dollar values that percent increases imply and ensuring that they are reasonable given the overall size of the market that the firm operates in.

Keep in mind that fast revenue growth is fleeting since strong performance attracts competition. High growth rates decay very quickly and regress to the long-run median of the industry. It is especially wise to be skeptical of strong growth among mature businesses since it is difficult for firms to sustain rapid growth as they expand drastically in size. Extremely large Fortune 50 companies struggle to grow in particular, averaging growth only 1% above inflation.

The following real revenue growth rates are taken from McKinsey’s text, Valuation (2005, p. 155). Although median revenue growth was 6.3% in real terms between 1963 and 2003, it fluctuated wildly between 1.8% and 10.8% on an annual basis. Aswath Damodaran additionally keeps a table of historical (nominal) growth rates by sector online (here).

5. Length of Short-Term Growth
When projecting future revenue growth, firms are often modeled as having a high rate of growth in the first period and then becoming stable thereafter. An analyst must therefore consider how long the short-term growth rate will last in addition to how high it will be. Though a common high-growth period is five years, the period length will largely depend on a thorough understanding of the industry and the competitive context of the company. It is valuable to consider more than one possible outcome.

6. Terminal Growth Rate
Analysts should avoid applying the same short-term growth rate to the long-term growth rate used in the terminal year. The long-term growth rate should generally not be greater than the growth rate of the market in which the firm operates. Therefore the best estimate of long-term growth is likely the expected long-run rate of consumption growth for the industry. This can often be approximated by looking at the sustainable growth rates of industry competitors (=ROE x reinvestment ratio) or referring to historical growth rates by industry group (above). Even then, the firm’s long-run rate will likely be lower than these estimates since the economy is always composed of high growth companies that push the average up. Another proxy includes simply using the 10-year treasury risk-free rate.

Concluding Thoughts
If a firm operates in multiple types of businesses, revenue should be forecasted by segment rather than on a consolidated basis. It is also not uncommon to show a company’s high growth rate converge gradually toward its long-term rate. Revenue might increase by, say, 10% but decline by 1% each year until reaching the industry median of 5%.

Revenue growth can be misleading due to changes in currency values for multinationals, mergers and acquisitions, and changes in accounting policies. Therefore when analyzing revenue it is worth breaking down growth into (1) organic growth, (2) currency effects, (3) acquisitions, and (4) accounting changes. It is additionally worth keeping in mind that revenue growth is driven by unit prices and quantities sold. If the data is available, more detailed forecasts can be created using this relationship. Capacity-based forecasts can also be modeled for, say, retailing where revenue is based on same-store sales growth and sales from new stores.

It is finally worth noting that analysts also provide estimates of growth for the companies that they follow. Companies with large market capitalizations and large institutional investor bases tend to have the most analysts following them. Analysts theoretically should provide better growth forecasts than models that rely solely on historical data since analysts closely monitor all publicly available information, incorporate data on the general economy, track competitors closely, and have access to private non-material information about the firms they follow.

While analysts do tend to provide superior projections in the short-term, there is little empirical evidence showing that analysts accurately forecast long-term growth rates. Empirical research also suggests that research analysts tend to be overly optimistic in longer forecasts, thereby suggesting that they might serve well as upper limits to growth forecasts. Services that report on these estimates generally provide analysts’ estimates of growth in earnings. Since revenue growth rates are commonly lower than growth rates in earnings, you will need to revise these estimates down if using analyst forecasts. Although analysts can make significant forecasting errors, their estimates should be incorporated into your valuation when there have been significant changes to the firm or industry in the recent past.