Engines of Value
- Buybacks, Earnings Growth, P/E Multiples and Stock Returns -
Introduction
We recently came across a lecture from John Huber (Saber Capital), where John breaks down stock returns its three components: earnings growth, change in P/E multiple, and the change in shares outstanding.[1] In his lecture, John uses the examples of Autozone and Apple to demonstrate how buybacks can add to investors’ returns. We want to build on that concept and see whether the data of the S&P constituents reveal any interesting patterns. How did S&P companies do over the last ten years, and what were the main contributing factors to those returns? Did stock buybacks result in higher returns for shareholders? Did the market penalize EPS growth purely driven by buybacks? Did low or high P/E multiples help to predict subsequent returns?
Definitions
We define the three engines as follows:
1) Buyback Engine = Shares Outstanding (at start date) / Shares Outstanding (at end date)
2) Earnings Engine = Earnings (LTM) (at end date) / Earnings (LTM) (at start date)
3) P/E Engine = P/E (LTM) (at end date) / P/E (LTM) (at start date)
The Buyback Engine ratio is greater than 1 for companies that bought back more shares than they issued over the time horizon and less than 1 for companies that diluted shareholders. The end date is June 30, 2023 and the start date is July 01, 2013.
The resulting stock price return is simply the product of the three engines:
Stock Return = Buyback Engine x Earnings Engine x P/E Engine
Data and Limitations
The data was obtained from S&P Capital IQ. To get meaningful results and summary statistics we eliminated data points with negative earnings or where pricing data at the respective start date was not available.[2] For the most part, we use median instead of average values, since the averages for the underlying data are highly impacted by extreme outliers.
The underlying analysis represents a snapshot in time and can therefore only serve as a rough indication. The results might differ substantially when looking at different time windows and respective start/end dates.
Analysis
With these background settings in mind let’s jump into the data…
The vast majority of stocks generated positive returns over the last 10 years, with a range of -11% to a stunning 60% per year (#AIHYPE) and a mode of around 6%. Only 42 out of 388 stocks had negative returns and equally 42 stocks had returns of around 20% or more. Quite a few companies were able to please investors with outsized returns of more than 30%, but more on those later.
Sector Valuations
The median and average P/E (LTM, 2013) across all sectors were 20x and 43x respectively. The average P/E drops to 23x after excluding extreme outliers.[3] Pretty much most sectors traded in line with the overall median, except for Real Estate (median of 47x). The median P/E for Energy, Financials, and Utilities were slightly on the “cheaper” end (all around 18x).
It is probably no surprise for most readers that stocks have been re-rated to higher P/E ratios over the last ten years. This effect is somewhat significant and added 4% per year for the S&P as an aggregate. The median and average P/E (LTM, 2023) across all sectors were 22x and 39x respectively. Interestingly, valuations for 2023 seem a lot more dispersed. Many sectors exhibit a much wider interquartile range (IQR) and most sector medians differ substantially from the overall median. Take Information Technology as an example. The IQR more than doubled from 15 in 2013 to 32 in 2023 and the median P/E moved from 20x to 29x respectively.
P/E Multiples and Subsequent Returns
Did low or high P/E multiples in 2013 help to predict subsequent 10-year performance? In short, not quite. The scatterplot below shows pretty much no relationship between P/E multiples in 2013 and the 10-year returns. The data points appear somewhat randomly distributed.
Although not shown here, the picture does not change if we use forward multiples instead of LTM.
What is interesting, however, is that if we account for quality, we can see that high-quality companies seem to outperform their lower-quality peers. Quality is defined as ROIC (return on invested capital, operating profit / total capital) greater than 10%. We use green and grey to separate different subgroups. Green relates to companies that generated an ROIC of more than 10%, and grey relates to those that had an ROIC of less than 10%. Far more companies that generated a high ROIC also achieved a 10-year return of more than 10%.
Growth in EPS vs. Growth in Earnings
It is often argued that investors prefer companies that can grow their bottom line by increasing sales or through cost savings. EPS growth by simply reducing the number of shares outstanding is considered of lower quality. We are not quite sure we fully agree with that view, but we do acknowledge that EPS growth through buybacks is theoretically limited, whereas increasing sales seems at least less limited.
The chart below plots the EPS growth against the earnings growth of our subset of the S&P. The dashed yellow line separates the companies into two groups: one group where EPS growth is higher than total earnings growth (area left to the yellow line) and one group where total earnings growth is higher than EPS growth (area right to the yellow line). The two plots zoom in for different levels of the respective engines, to make any relationship more visible to the reader. So, how did the subset perform?
Green represents those companies that achieved an annual return per year of more than 10% and grey for those that had returns of less than 10%. If the market would penalize EPS growth purely driven by buybacks, we would see more green dots to the right of the yellow line. This is not the case. Rather all companies of our subset fall into one of the two categories without any systematic pattern.
The Buyback Engine in Isolation
If we now turn our analysis to the buyback engine we can see that not all buybacks seem to “create value”. The following chart groups all stocks into bins, according to their percentage of buybacks. The number on top of each column represents the median return of the respective group/bin.
Most companies effectively did not buy back any shares (see 1.0x bin), but did relatively well in terms of median return. It looks like companies that reduced their share count did slightly better than their diluting counterparts (comparing the median returns left to the 1.0x bin to the ones right to the 1.0x bin). The data however is somewhat inconclusive, since some diluting companies also generated relatively decent returns (e.g. 10.5% for the 0.6x bin), and some companies that engaged in massive buyback programs had rather poor returns (greater 2.0x bins). Not apparent from the chart, but worth mentioning, is that from those companies that reduced their share count by at least 50% roughly half saw their stock price increase by more than 10% per year, whereas only 17% of those that diluted their shareholders by at least 50% were able to generate more than 10% per year.
We are tempted to conclude that buybacks add favorably to stock returns, even though there are positive and negative exceptions.
The Three Engines of Value
Let’s see how the combination of the three engines played out as a whole and also by sector. The median buyback engine across all companies added 1% per year to the overall return, roughly 2% for the P/E engine, and 8% for the earnings engine. The dominant contribution of earnings growth is hardly a surprise, the rather low contribution of P/E re-ratings probably is.
Consistent with our conclusion from above, the three sectors that diluted shareholders (Energy, Real Estate, and Utilities) experienced the lowest median returns. In the case of the Energy sector, pretty decent earnings growth did not help to prevent a massive downrating for most stocks. The five sectors that performed best had their earnings engine as the major source of their returns (i.e. at least 7% contribution). Two of those sectors had all three engines work in their favor (Information Technology and Industrials).
Top Performing Stocks
Ten stocks in our group of companies literally smashed it, generating returns greater than 30% per year, for 10 years! That’s an amazing performance. No surprise, Nvidia is one of them. Notably, 7 out of ten are tech stocks, with five being semiconductor (or semi-related) companies. Even though the semiconductor industry is currently emerging from a downturn, the earnings growth for most companies in the industry since 2013 is quite remarkable. Also, quite interesting, only two out of ten had all three engines work in their favor (Fair Isaac and MSCI), and five out of ten actually diluted shareholders (i.e. buyback engine less than 1x).
Worst Performing Stocks
Let's shift our focus to the stocks that have shown the weakest performance. Interestingly, in line with the best-performing cohort, five out of ten of the worst performers diluted their shareholders but did so at a slightly larger scale (median of 0.9x and average of 0.8x compared to 1.0x). On top, earnings growth was a lot weaker, and the majority of stocks experienced quite a significant downrating in their P/E multiple.
SEC Amendments to Modernize Share Repurchase Disclosure
On May 3, 2023, the SEC announced to make significant changes to the disclosure around buybacks. The SEC hopes to “increase the transparency and integrity” and help investors to “better assess issuer buyback programs”. For quarters beginning on or after October 1, 2023, companies must disclose daily buyback volumes, the average price paid, and whether certain officers and directors traded in the securities four days before or after the buyback program was announced. More interestingly, companies are also required to disclose the objectives or rationales along with any policies relating to purchases and sales of their securities.
We do welcome the additional disclosure, given the increased scale of buybacks over the last few years. Investors will be able to gain more insight at what levels companies buy back and sell their shares. In our present study, we were not able to analyze whether buybacks at more reasonable levels outperformed. With the amendments by the SEC we can have a deeper look at the topic.
Conclusion
The single most important factor for stock returns in the S&P over the last 10 years was earnings growth, even though decent earnings growth does not help if the other two engines do not work in your favor (e.g. Energy sector). Buybacks do not single-handedly produce magical returns but can be a pleasant support. It seems that heavily diluting companies fare worse, compared to companies that reduce their share count or keep their share count even.
Systematically screening low or high P/E ratios won’t help produce outsized returns. Some companies that look “cheap”, are actually cheap for a reason. Others that look expensive do not live up to the expected potential. When accounting for quality the picture does change. High-quality companies that produce high returns on invested capital did achieve significantly higher returns.
We hope you like our post and don’t forget to subscribe. Feel free to send us any comments and questions at analyst@catalytics-research.com.
Footnotes
[1] “3 Engines of Value: Guest Lecture at NCSU (3/29/23)”, available at: https://sabercapitalmgt.com/letters-and-notes/.
[2] A total of 112 data points have been eliminated, mostly due to the positive earnings restriction.
[3] Excluding extreme outliers with a P/E greater than 100x eliminates 12 data points.
Disclaimer
The information provided in this article is not considered investment advice and is for educational purposes only. Do your own due diligence!