Some variation of “Is value investing dead?” has been one of the chronically recurring headline/discussion topic in the past decade among investors (especially value investors). From a definitional perspective, value investing is always going to work – after all, acquiring any economic asset at a cost that is substantially less than what it is worth is almost always going to be profitable.
That said, when financial media is agonizing over the “death of value investing” it is typically referring to (systematic) value investing based on some set of valuation metrics (P/E, P/B, P/S, EV/EBIT, etc.). Obviously, no strategy works all the time. Also, in our view, with the scale of computing (and intellect) chasing excess returns, previously profitable approaches/ factors are likely to degrade over time¹. As such, investors will just have to find new ways to winnow out stocks that are cheaper for a reason from those that are cheaper because of investor overreaction.
This post looks at the performance of High-Quality Systematic Value Investing² primarily as a means to discover patterns and incorporate those lessons to improve our investment process – and to a certain extent to also satisfy our curiosity about returns from Magic Formula³ /Quantitative Value⁴-based approaches.
TL;DR:
No strategy works all the time and past performance is not indicative of future returns
Even more importantly, a given strategy may not be suitable for you/your clients – we have different temperaments, time horizons, risk appetites, return expectations, portfolio preferences, etc. It is better to adopt a sub-optimal strategy that you can stick with than a theoretically better strategy that you may be highly likely to disband when it goes against your expectation
Relative valuation will not adequately protect an investor from overpaying. When pursuing systematic investing, one needs to be especially wary about cyclical firms and commodity firms that often tend to have inflated earnings yields at the outset of a contraction after a long business cycle expansion and/or commodity cycle peak
Purely quantitative approaches can often miss serious qualitative risks (for e.g. business model weaknesses, likely outcome of legal issues, loss of exclusivity, etc.). As such, in addition to quantitative factors, it is important to incorporate checks for qualitative risks in the investment process
No matter how good our skill, we are bound to make errors – it can be very helpful to practice patience and pay careful attention to the balance sheet as it can often (but not always) help overcome errors in judgment/uncertainties that come out of nowhere
The Setup
Value investing as a philosophy resonates naturally with my mostly cautious and patient temperament. My background is in management consulting, analytics, and (software) engineering and I initially conducted backtests for 2003 through 2012 as a means to accelerate learning, identify likely risks/pitfalls, and help improve my investment process.
Since then, the approach detailed below has been run at the end of each year (during the initial years also at each quarter-end) to assess how the model-recommended portfolio would perform and glean lessons for further improving the investment process. Here is a brief look at the model implemented at our end (for more info on specific aspects of the model, please refer to footnotes for books/papers):
Exchange | US-listed stocks domiciled in English-speaking countries excluding ADRs and OTC stocks. |
Market Cap | > $50M |
Relative Valuation | Cheapest 10% of firms based on EBIT/EV |
Quality Factor | Top 50% of firms based on a composite score formulated using ROA, ROIC, CFOA, and Gross Margin of past 7 years together with Piotroski Score largely based on Quantitative Value⁴ approach |
Sector exclusions | No financials, REITs, utilities (to avoid highly-leveraged sectors) and ore extraction (personal preference) |
Earnings Accruals | Exclude firms with high scaled accruals - i.e. those in top 5 percentile of Sloan ratio⁵ |
Bloated Assets | Exclude firms in top 5 percentile of SNOA metric⁶ |
Fraud risk | Exclude firms with probability of manipulation > 0.95 based on Beneish PROBM model⁷ |
Bankruptcy risk | Exclude firms with Altman Z-Score⁸ <= 1.81 |
Debt/Equity | Exclude firms with Debt/Equity >= 75%, as we wanted to avoid firms with high leverage |
Debt Repayment | Exclude firms that would require > 5 years to payoff debt with current FCF to avoid firms with insufficient debt servicing capacity |
Share Dilution | Exclude firms with share dilution > 10% |
FCF Decline | Exclude firms with steep decline (> 25%) in FCF or negative FCF |
Revenue Growth | Exclude firms that did not show any revenue growth in 3 years |
Valuation | Exclude firms with price above estimated intrinsic value calculated primarily based on the Earnings Power Value approach with a set of heuristically selected parameters |
The final six filters are based on patterns observed in our 2003-2012 backtests. We included these because, in our view, when utilizing a purely static model lacking the intervention of human judgment, it is prudent to err on the side of caution.
The model portfolio is constructed by ordering output from the model using the composite quality score discussed above and then choosing the Top 20 from the list.
The Results:
The slideshow below presents the comparative returns from 2013 through 2020 and also the portfolio components selected by the model for each of those years.
Key Lessons:
For brevity, the Results section above presented only annual returns from 2013-2020. However, the key lessons come from our observation of model portfolio components from 2003 onwards - especially those that had steep declines (>20%) in a given year and those that suffered a prolonged/unrecovered decline.
Relative valuation will not adequately protect an investor from overpaying. This is especially true about cyclical firms and commodity firms that often tend to have inflated earnings yields at the outset of a contraction after a long business cycle expansion and/or commodity cycle peak. For example, many of the stocks that saw the steepest declines in 2017 and 2018 (see the previous section) were in highly cyclical industries such as Semiconductors, Autos, and Apparels.
Leverage can seriously exacerbate business decline - especially excessive hidden leverage (e.g. off-balance sheet operating leases). The retail apocalypse seen in the previous few years is a good example of this (incidentally several retail firms showed up on Magic Formula/Quantitative Value-based screens in 2017 & 2018). What is precarious about this hidden leverage is that it also inflates return metrics (such as ROA, ROIC, CFOA, etc.) thereby giving them a higher perceived quality than warranted. For example, EXPR (an apparel company, that was part of the model portfolio in 2017) should not even have been part of the model portfolio as accounting for operating leases its Debt/Equity would have been ~150% (vs 0% reported) and even more importantly its ROIC would have been ~10% (vs ~30% per traditional definition), which would have severely impacted its quality score. Others, such as FRAN, were in a worse situation and eventually went bankrupt (FRAN should also not have made the portfolio as it was also highly levered). Fortunately, the adoption of FASB's new lease accounting standard (ASC 842) is helping reflect true leverage on balance sheets.
“The big money is not in the buying and selling … but in the waiting.” – Charlie Munger. Business cycle/commodity cycle/corporate turnarounds are not going to align smoothly with a model's annual rebalance timeframe. When investing in cyclical businesses and/or businesses that require turnaround - it is important to consider both our ability to understand as well as our and the firms' ability to withstand the length of cyclical/business downturn. Patience and a sound balance sheet can be strong allies in enabling profitable investments. For example, after a few trying years, many of the retail/footwear names that had sound balance sheets (WSM, HIBB, DKS, BKE to name a few) rebounded strongly after the Amazon-driven retail disruption fears abated. Similarly, after a couple of years (and steep declines which provided even better buying opportunities) most of the Semiconductor and Auto firms that came up in the above model screens also recovered very profitably.
Even exceptional companies with phenomenal ROIC and long growth runaways do come on sale occasionally. For example, AAPL in 2013, 2016, and FB in 2018, 2019. Please note that what we mean here is that they got so cheap as to appear in the Top 10% or Top 20% of cheapest firms by earnings yield (though, their availability at a significant discount to their respective intrinsic values was not as infrequent).
Purely quantitative approaches can miss serious qualitative risks – as such, it is essential to incorporate a step to consider qualitative risks in our investment process. It is important to note that not every risk is going to materialize nor will it necessarily be material - however, being aware of the risk can help the investor judiciously determine the appropriate margin of safety or avoid the investment entirely in those cases where accretive risks substantially increase the likelihood of capital loss. Some of the recurring patterns we observed included: - Customer Concentration Risk - Product/Service Concentration Risk - Supplier Concentration Risk - Geographic Concentration Risk - relevant for businesses in unstable countries - Legal Outcome Risks - One-time/temporary business benefits Lannett Co (LCI) is a rather egregious example of cumulative qualitative risks – in 2015, it had 75% gross margin, ~35% ROE, and ~24% FCF/Sales but has since lost ~90% of its market cap. Many value and quant investors (and also the model above) missed LCI's highly compromised business model (concentration risks in customer, product, and supplier) by focusing on quant metrics that looked great then. However, LCI's business precipitously declined when it could not renew its distribution agreement with Jerome Stevens Pharmaceuticals for Levothyroxine Sodium tablets, which accounted for more than 50% of its revenues.
A purely formulaic approach can be very difficult to adopt (at least it is for us, but to each her/his own). Even more importantly, sticking it out through volatility can be really difficult if you are unsure of what you have invested in. For example, Stamps.com (STMP), a shipping solutions technology provider started coming up on model outputs in Q1, 2019. STMP had ~80% gross margin, ~30% ROE, and ~45% FCF/Sales - despite this, by mid-2019 it had declined by ~75% YTD (from ~$155 to ~$35) and ended the year down by nearly 50%. In our view, not many would have the fortitude to continue holding on after such a steep decline (especially in a year where the S&P500 returned ~30%) – unless they understood the business and recognized the overreaction. STMP is also a good example to contrast with LCI above. In the case of LCI, it would have been prudent to avoid the investment altogether given multiple concentration risks. However, in the case of STMP – despite the overwhelming dependence on USPS, the primary revenue segment at risk was < 5% of total revenues. And so being aware of the risk, would have helped in determining and patiently waiting for an appropriate margin of safety. In our view, even with complete loss of that segment and the expected decline in operating margin in the near-term (primarily to fund investments to help diversify sources of revenue), anywhere below $60 per share, the risk of loss would have been almost non-existent (STMP was acquired for $330 per share by Thomas Bravo in September 2021).
Temperament and Time provide two of the most effective edges in investing – paraphrasing Jack Treynor: one is more likely to benefit from market inefficiency in 'slow-moving ideas' – ideas that require reflection, judgment, and special expertise.
Moving Forward:
The logical next step would be to incorporate the lessons learned above - and that's exactly what we did when we launched Cognitive Quant platform in February, 2020. We leveraged advances in Business Reporting Standards, Natural Language Processing and AI to address many of the cognitive, informational, and analytics aspects identified in the prior section. While being cognizant that it is a single period instance, you can read about the year-end performance of the improved version⁹ here.
Invert, always invert. — Carl Jacobi
Carl Jacobi's maxim on inversion (advocated by Charlie Munger) is a great mechanism to improve results by consistently minimizing mistakes. As such, our preferred approach is to use the refined model primarily to generate a list of investment candidates. We then utilize checklists, qualitative risk insights, valuation framework and other features in the platform to more thoughtfully consider the downside risks and thus improve the investment process.
Subscribe to access all the platform features and see for yourself how it can help improve your investment process.
¹ For example. we would be hard-pressed to find Net Net Working Capital stocks in developed markets except during significant market dislocations.
² Value as determined by cheapest decile based on EBIT/EV; Quality defined as stocks that are above average in terms of a composite quality score based on historical performance across business cycle of the firm's ROA, ROIC, CFOA, and Piotroski Score among other metrics largely based on the Quantitative Value approach (see Setup section for details).
³ Greenblatt, Joel et al (2010) The Little Book That Still Beats the Market
⁴ Gray, Wesley R. et al (2012). Quantitative Value
⁵ Sloan, R. G. (1996). Do Stock Prices Fully Reflect Information in Accruals and Cash Flows about Future Earnings?
⁶ Hirshleifer, David A. et al (2004). Do Investors Overvalue Firms with Bloated Balance Sheets?
⁷ Beneish, Messod D. (1999). The Detection of Earnings Manipulation
⁸ Altman, Edward I. (2000). Predicting Financial Distress Of Companies
⁹ Please note that the improved version is not a like-for-like comparison - the referenced post describes the setup
