V. An Evidence-Based Approach
Watch Our Video on Evidence-based Investing
Key take-aways
- Empirical evidence strongly supports the efficacy of valuation and profitability disciplines as factors for return premiums, as well as small cap.
- As time frames lengthen, the probability of realizing these premiums increases significantly. The premiums have been persistent over time.
- The value and profitability disciplines (factors) are complementary and best viewed as an integrated investment approach.
- The value-profitability discipline is efficient and cost effective to implement and best pursued systematically to maximize reliability.
Let’s turn now to the hard numbers. I present empirical evidence from 90 years of stock market data. This data supports the conclusion that adhering to the sensible disciplines discussed in the last section results, over time, in excess returns above the broad market. I believe a key feature for any investment fund is reliability—the fund will likely endure and deliver upon its stated objectives. An investment strategy is more reliable when (1) there is strong evidence of its efficacy and (2) it can be implemented systematically. Here we check both boxes.
Discipline as Factors
The graphic above summarizes the empirical case for the excess returns available above the market return. Using data from the public website of Prof Kenneth French, a leading finance scholar at Dartmouth, three measurable stock characteristics—size, valuation, and profitability—have each generated premiums. Going back to 1927, the premiums for small stocks (smallest 30% of the market) and cheap stocks (cheapest 30% of the market) averaged 2.2% and 3.1% per year, respectively, on top of the 9.8% total market return. Going back to 1964, the premium for stocks of profitable companies (top 30%) averaged 1.2%, above the 10.0% total market return. These are significant premiums, which result in large differences in compounded wealth over time.
This is why I think it makes sense to make a distinction between large cap and small cap companies (in addition to the higher price volatility of small cap). And this is also why I think it makes sense to adhere to a valuation and profitability discipline for the stocks of companies of all sizes.
As highlighted in the section on Index Investing, academic study after academic study has reached the same conclusion. Where I refer to “disciplines,” the word I feel best captures the nature of the exercise, academics refer to “factors.” This is because they use these factors mathematically in the context of pricing models for stocks. These models are meant to estimate the expected return required for a stock or portfolio of stocks based on its characteristics—including size, valuation multiple and profitability. Prof French and his colleague Prof Eugene Fama have published the most well-known pricing models. Their most recent model, known as the “5-factor model”, is shown at left.
Criteria for Robustness
It is important to consider whether these premiums are just statistical flukes. Statistical tests of significance indicate it is highly unlikely these premiums occurred at random. Moreover, additional considerations speak to their robustness. The observations persist over time, which we discuss in more detail just below. They hold across geographic markets (US and overseas stock markets). They also work if the methodology is tweaked. For example, using various valuation measures (whether price to book, price to earnings or price to free cash flow) in studying the value premium. Or for all measures, whether the universe is broken into deciles, quintiles or quartiles. Finally, and most importantly, we can identify the common sense reasons that give rise to the premiums, which we discussed at length in A Sensible Approach.
The Premiums and Time Frame
Investment returns are unpredictable from year to year. This is true of overall stock market returns, and it is true for factor premiums as well. You can see this in the table immediately below, which looks at all the rolling periods (1-year, 5-year, 10-year, 15-year, 20-year) from 1927 to 2016 (starting 1964 for profitability).
Table 3: “Batting Average” of Premiums over Different Time Periods | |||||
---|---|---|---|---|---|
20 Years | 15 Years | 10 Years | 5 Years | 1 Year | |
Size (Small Cap) Premium | 86% | 76% | 70% | 65% | 54% |
Value (Low Multiple) Premium | 99% | 92% | 85% | 74% | 61% |
Profitability (High RoC) Premium | 100% | 95% | 86% | 78% | 68% |
Source: Historical Benchmark Returns; Data Library of Kenneth R. French, Professor of Finance at the Tuck School of Business at Dartmouth College.
As you can see, whether small cap stocks deliver a premium in any given year is not much more than a coin toss, though the value and profitability premiums show up a bit more often. But as time frame extends, the probability of realizing the premiums grows significantly. This is what I mean when I talk about the persistence of the premiums over time. Based on historical data, with 20 years of investment horizon, the value and profitability premium materialized in nearly every period, and the size premium materialized in a robust 86% of periods.
This statistical study of premiums in the context of time frame is a bedrock piece of analysis for Constancy Investors. It should inform any investor about how to use funds that seek to harvest these premiums in your portfolio. As the time frame for your goals shortens, the probability of realizing the return premiums we target decreases, though they remain good. Accordingly, for portfolios designed for shorter time frames, I’d suggest paring back allocations to strategies designed to capture the premiums, small cap in particular, in favor of low cost, large cap funds that closely mirror the S&P 500, the index most commonly identified with the US stock market.
The Intersection of Value and Profitability
The table at right shows a matrix of the compound average annual returns for small cap companies from 1964 to 2016, divided into quartiles for both the valuation and profitability factors. Each of the 16 “portfolios” in the matrix is market cap weighted. Note that the returns go steadily higher whether you move up any column or across any row.
For any profitability quartile, returns improve as stocks trade at cheaper valuation multiples. And for any valuation quartile, returns improve as companies generate higher profitability (returns on capital). This strong correlation across quartiles is further evidence of statistical robustness. Note that this matrix is for small cap because it is so striking; large cap exhibits the same pattern, though the correlation and differences in returns are not quite so dramatic.
In addition to being significant in their own right in delivering premiums, the valuation and profitability factors (or disciplines) have an added bonus. They are quite complementary. Controlling for valuation improves the returns of profitable companies, and vice versa. This is again evident in the pattern of returns through the matrix.
Moving diagonally from the southwest to the northeast in the table, returns climb as profitability increases and valuations grow less expensive. Additionally, as the prices of profitable and value stocks tend to vary in how their prices move, using the intersection of the two factors helps reduce price volatility.
The numbers are the numbers. But what again makes the statistics more compelling is the common sense intuition underlying the intersection of profitability and valuation. We cannot look at price in a vacuum.
To use an analogy, price alone is insufficient to think about whether you are getting a good deal on a new car. To do so, you need to think about price relative to the make and model and features of the car. A bargain price on a baseline Ford Focus is going to look much different than a bargain price on a fully-loaded Mercedes S-Class.
The ultimate attribute we care about for companies we are buying is the profits they generate and how efficiently and dependably they generate them. So, in thinking about whether we are getting a good deal on an investment in a company’s stock, I think it is sensible to look at valuation relative to profitability. The empirical evidence—the statistics—again support this underlying logic.
As such, in my view, the value and profitability factors are best thought of as an integrated approach. Very simply, to construct portfolios of companies with the best mix of value and profitability attributes.
Systematic Implementation
The many empirical studies undertaken by academics all adhere to a tightly defined methodology. Given the objective is to capture the return premiums supported by the empirical studies, a good question is whether these methodologies can be substantially implemented in practice to construct real-life portfolios.
The good news is that the value-profitability intersection can be implemented quite efficiently in practice. Below is a graphical snapshot of such an approach for US large cap stocks. I show large cap simply because it results in a cleaner picture, as it has less companies. (Note that dividing the US stock market into small and large cap is extremely straightforward). The size of the circles reflect the market cap (size) of the company, and it’s position shows how it ranks with respect to both valuation and profitability. Just as the table above suggests, systematic fund managers can construct a portfolio by eliminating entirely those companies in the “southwest” (weak profitability and expensive valuation) and skewing the portfolio toward those companies that sit in the “northeast.”
Portfolio construction in such a manner proves to have many advantages in practice. First, portfolios are quite diversified, with dozens and dozens or even hundreds of stocks. The number of holdings can be tailored for diversification preferences and trading costs, which may vary among investors. The return-volatility profile will differ among more or less diversified portfolios, but factor premiums hold over time.
Low turnover helps limit ongoing transaction costs. The picture above reflects the large cap market in late 2017. Over time, the circles (i.e. companies) will move around on the grid, as either their valuation multiples or profitability change. But, based on historical data, such movements are gradual. A company’s profitability has been quite persistent over time, so valuation changes are the predominant driver for turnover. And this is good. Required sells will likely result from higher valuations, which means the stock has been moving up in price. This momentum and improved liquidity facilitate sales at attractive prices and low implicit trading costs (i.e. not crossing the bid-ask spread).
A systematic approach also introduces great flexibility in portfolio construction. The objective is for the overall portfolio to continuously reflect attributes with respect to valuation and profitability on average across the portfolio. In mathematical parlance (as in the five-factor model shown above), for the overall portfolio to reflect a targeted “exposure” to the factors. A systematic fund manager can achieve this targeted exposure by selecting from a wide variety of companies available in the universe and by holding them in varying sizes for each position. As a result, a systematic fund can be highly selective in what it chooses to buy and sell and when. At any moment in time, it can transact only in those stocks trading at attractive prices. This same flexibility has an added bonus for taxable accounts. It allows for a great degree of latitude in managing to minimize realized capital gains, including active tax loss harvesting.
So, in short, systematic implementation is not only doable, but has fringe benefits. I also strongly believe that systematic implementation is the best approach. It cannot be improved upon. Very simply, the strong evidence for the return premiums is the basis for the strategy. The approach in implementing the strategy should thus adhere to the methodologies used in the empirical studies as closely as practicable.
Moreover, the evidence for the value of active management beyond doing this, for example in single stock picking or in timing the market, is quite poor. Such activities at the end of the day rely on qualitative judgments and forecasts, which Graham and Dodd rightly de-emphasized, as discussed in A Sensible Approach. Recent findings in behavioral finance, which I introduced in Recognizing Your Emotions, show their wisdom in doing so. We now know that all of us too often arrive at incorrect judgments and poor decision-making, simply due to how our brains are wired.
Given this and their historically high fees, perhaps it is not surprising that less than 20% of stock mutual funds outperformed their benchmark for the fifteen years through 2016. Only about 50% even survived. This is problematic for an inherently long-term activity like investing. And why we emphasize the reliability of an investment strategy so strongly. A systematic approach, strongly supported by empirical evidence, is the most reliable investment approach.
To sum up, here are the common sense ideas in support of An Evidence-Based Approach to investing and why it maximizes the likelihood of success:
- construction of portfolios through adherence to sensible disciplines
- robust empirical evidence, using the scientific method, supports the efficacy of the disciplines as factors in higher returns
- basis for portfolio construction is more soundly-reasoned than index methodologies (and thus aping such indices)
- systematic implementation, so as to adhere to the methods of the empirical studies, that is viable and cost effective
- reliable and enduring execution, as rules-based and not dependent on the skill of a single individual
Under the Invest tab above, I provide examples of funds that adhere to this evidence-based investment approach–funds managed systematically to implement these disciplines–and later show how they can be used in a portfolio context.