The Persistence and Predictability of Growth
We recently wrote about the lottery-like characteristics of small-cap growth stocks: the majority of these companies lose money while a tiny few have lottery-like returns. Why is this the case? And how should investors change their behavior in light of this data?
We define growth stocks as companies that meet two criteria: 1) historical growth rates of revenue and profits above the market average and 2) high valuation multiples implying above-market growth expectations. The latter criteria would be the factor leading to the disappointing results of investing in growth stocks.
We believe that betting on growth is a strategy bound to disappoint because growth is neither persistent nor predictable.
The best and most in-depth review of this topic is Louis Chan, Jason Karceski and Josef Lakonishok’s paper The Level and Persistence of Growth Rates, a staple read for equity investors. The researchers looked at the U.S. stock market from 1951–1997 and examined how persistent revenue and profit growth were, and how predictive analyst estimates and valuation ratios were of future growth.
Upon reviewing nearly 50 years of data, their conclusions indicated a lack of persistence in long-term earnings growth. Neither analyst forecasts nor valuation ratios have significant predictive power. The paper argues that investors over the period would have been more accurate using GDP growth for every forecast than making individual predictions based on analyst estimates or on growth rates implied by valuation ratios.
This is at odds with the consensus method of active managers, most of whom use discounted cash flow models to value stocks. “Market valuations reflect a pervasive belief among market participants that firms who can consistently achieve high earnings growth over many years are identifiable ex ante,” wrote Chan, Karceski and Lakonishok. Despite widespread belief that multiples signal future growth rates, they find that “the cross-sectional relation between earnings yields and future growth is weak.” In other words, paying high multiples for businesses because you believe they will grow faster than the market is a not a smart decision.
The evidence that the researchers display for their conclusions is presented below.
First, they attempt to show that there is almost no persistence to growth. While above-average revenue growth seems to show some mild amount of persistence, that persistence appears to vanish once you get to profitability — the surer companion of valuation. The number of companies that have streaks of above-average growth is roughly what you’d expect from pure chance. You can see this by looking at the line below labeled “percent above the median” that looks roughly like 50%, 25%, 12.5%, etc. Within a few years, it is almost statistically impossible to discern future growth candidates by looking at those of the past. This can be particularly problematic given that most prevailing valuation methodologies derive 80–95% of their value from earnings in years beyond the 2–3 year horizon (the so-called terminal value).
Figure 1: The Persistence of Growth Rates of Operating Performance
Source: Chan et al.
Nor are there subsets of firms that display persistence. While many market observers hypothesize that technology or pharmaceuticals companies should display persistent above-market growth, this may not actually be the case. Nor do large companies, small companies, glamour stocks, or value stocks show any persistence of above-market growth.
Figure 2: Persistence of Growth Rates for Selected Classes of Equities
Source: Chan et al.
The authors try virtually every other permutation of stock categorization and still find similar results on growth persistence.
It is notable that even in the U.S. technology sector, where consensus forecasts by Wall Street analysts and research shops such as Gartner and IDC are often way above market-wide company growth levels, there seems to be almost no ability on an individual firm level to pick income statement outperformers against the broader market. It may be that “mega-trends” of growth in themes like big-data, hyper-converged infrastructure, virtual reality, and social media “app-ification” exist, but there is no statistical evidence we can exploit this via security selection predicated on future firm-level growth assumptions.
Each year brings a new trend, a new consensus forecast, a new view of what the future might hold. But these views are ephemeral. The prophets of profit would do well to heed C.S. Lewis’ advice: “the Future is, of all things, the thing least like eternity. It is the most temporal part of time — for the Past is frozen and no longer flows, and the Present is all lit up with eternal rays.”
The ex-ante predictive power of both growth rates implied by present valuation ratios and analyst forecasts also show little predictive power.
When it comes to growth rates implied by current valuation ratios, the authors find that “market valuation ratios have little ability to sort out firms with high future growth from firms with low growth,” and, furthermore, “Instead, in line with extrapolative expectations hypothesis, investors tend to key on past growth.” In other words, they believe that valuation ratios just tell us which companies have grown in the past, but have no predictive power.
And when it comes to analyst forecasts, “The dispersion in analysts’ forecasts indicates their willingness to distinguish boldly between high- and low-growth prospects,” they write. “Over long horizons, however, there is little forecastability in earnings, and analysts’ estimates tend to be too optimistic.”
So what should investors do? It’s simple: We believe that if the market valuation ratios and expert forecasts have no informational content for predicting future growth, then you would likely do better by buying the cheapest companies and avoiding companies that have to grow in order to justify their market valuations.
If you want to do any forecast modeling, you’d likely be better off using base rates such as GDP growth for earnings, rather than trying to forecast on the individual company level: This may lead you to overweight companies that are priced for little to no growth and underweight companies with high expectations. We believe that if you bet on unpredictability you would be right possibly more often than betting on a predictable future.