• Gaurang Merani

Examining Greenblatt’s “How the small investor can beat the market”


Introduction


The focus of this post is the research paper “How the small investor can beat the market” by Joel M. Greenblatt, Richard Pzena[i] and Bruce L. Newberg published in the Journal of Portfolio Management (1981). The paper was their Master's thesis while studying at Wharton business school. The objective of the paper was to largely examine the performance of securities that were trading at or below “liquidation value” during the 6 year period from April 1972 to April 1978 in the US.


Our objective was to analyze the study itself; determine its reliability, draw our own conclusions and glean, if any, actionable advice for the practitioner of the sub liquidation value method of investing.


Previous posts analysing the research related to securities trading at or below liquidation value include:


  1. An Analysis of “Benjamin Graham’s Net Current Asset Values: A Performance Update”

  2. An Analysis of “Graham’s Net-Nets: Outdated or Outstanding?”

  3. An Analysis of “Testing Benjamin Graham’s Net Current Asset Value Strategy in London”

  4. Analyzing Deep Value in the Eurozone

  5. What Has Worked in Investing (Tweedy, Browne) – Examining Net Nets

  6. Examining Saudi Arabian Net Nets



Key Methodology

• Valuation metric and portfolio construction parameters: “Liquidating Value Per Share” = (Current Assets – Current Liabilities – Long Term Liabilities – Preferred Stock) / Number of shares outstanding The authors measured the returns of four portfolios with their specific parameters identified below: Portfolio 1: Price/liquidation value: ≤ 1.0; P/E: floating with bond yields; Dividends: no dividend requirements. Portfolio 2: Price/liquidation value: ≤ 0.85; PPE: floating with bond yields; Dividends: no dividend requirements. Portfolio 3: Price/liquidation value: ≤ 1.0 P/E: ≤ 5.0; Dividends: no dividend requirements. Portfolio 4: Price/liquidation value ≤ 0.85; P/E: ≤ 5.0; Dividends: no dividend requirements. It is important to note that the sample they studied “did not consider stocks that had shown a loss over the preceding 12 months”.


• Weighting of holdings: Equal weight • Formation date: “we selected 15 segments of 4 months each over a six-year period.” This specification appears incorrect – the study extends for 6 years (i.e. 72 months) which would consist of 18 segments of 4 months (72 months / 4 = 18 segments). It appears that the authors created “overlapping portfolios” which would have reduced the impact of “timing luck” i.e. choosing a formation that that coincided with a market bottom (top) thereby potentially biasing returns upwards (downwards). • Sell strategy/Holding period: “We sold a stock after a 100% gain or after 2 years, whichever came first.” It should be noted that no mention was made as to whether the capital was immediately reinvested when a stock generated a 100% gain. However, we assume that portion of the portfolio remained in cash until the end of the 2 year holding period. • The stock universe: “We attempted to select a statistically significant and unbiased sample of stocks by compiling data on all firms listed in the Standard & Poor’s stock guide with market values over $3 million[ii] beginning with the letter A or B. This sample represented approximately 15% of all stocks listed in the guide, or 750 companies.” Presumably the Standard & Poor’s stock guide contained firms traded over the counter (OTC) as well as those listed on what was then known as the American Stock Exchange (AMEX) and the New York Stock Exchange (NYSE). For clarity it should also be noted that until the mid-1980’s the “NASDAQ” was synonymous with “OTC”[iii].



Reliability


While there are numerous biases/errors that can be made when conducting studies/back tests, below we have analysed those we deem most likely to impact a study of this nature: 1. Survivorship bias No mention was made with regard to controlling for survivorship bias. However, given the authors used the Standard & Poor’s stock guide (as opposed to a computerized data base that might not include delisted stocks) the data source would logically be free from survivorship bias. 2. Look Ahead bias Controlling for look ahead bias was not specifically mentioned. We assume: • Portfolios formed in April were based on data from the preceding December • Portfolios formed in August were based on data from the preceding June • Portfolios formed in December were based on data from the preceding September Nonetheless, in the absence of an express statement addressing look ahead bias it remains a possibility. 3. Time period bias The study spans 6 years from April 1972 to April 1978. We classify the study period as an “inadequate/unreliable”. For reference: · < 10 years; inadequate/unreliable · 11 to 20 years; somewhat reliable · > 20 years; more reliable · > 40 years; most reliable 4. Human error Given this study was created in a rather manual and laborious manner where the researchers had to manually collect the necessary data, we think it prudent to assume some errors may have occurred. 5. Journal rating/credibility [iv] While home to the research of many prominent academics and practitioners the Journal of Portfolio Management is not, to our knowledge, classified as a “top tier academic journal” and therefore cannot be granted the “additional credibility” that may come with such publication. Reliability Assessment: Notwithstanding the short time period (6 years) and a small sample size the study appears to be reliable.


Results and Analysis


For nostalgic reference the original “Table 1” summarizing the portfolio returns is reproduced below, however, given the study was published in 1981 type face we have also re-entered the data and presented it in an easier to read format:


If you immediately and intuitively understand the data in Table 1 we refer you to the following: “I know you think you understand what you thought I said, but I'm not sure you realize that what you heard is not what I meant” – attributed to Alan Greenspan A few points with regard to the figures stipulated in Table 1: 1. Period Returns - the returns presented in the body of the table represent the “4-month percentage increase in the portfolio”. These returns do not reconcile to the “Annual Compound Return” specified at the foot of the table. Indeed, each 4 month return represents a separate portfolio (inferred from the general variance in the number of holdings from adjacent periods). 2. Annual Compound Returns – we assume the Annual Compound Returns were calculated by applying the stated Sell strategy/Holding period whereby the authors “sold a stock after a 100% gain or after 2 years, whichever came first”. It is also stated that returns were calculated by “by assuming an equal dollar-weighted amount invested in each stock in the portfolio. Therefore, the percentage gain for the entire portfolio was merely an average of the percentage gain of the individual stocks”. Critically “Compounded Returns” reflect practitioner reality. 3. Sample Size - the authors “use[d] only 15% of the S&P stock universe […] there appears to be an opportunity to obtain a diversified portfolio of between 50 and 350 stocks that meet the constraints of Portfolio 1. In following our constraints, incidentally, we were unable to find any “bargains” in the market peak period of April 1972 to April 1973.” Nonetheless, the average portfolio contained 15 stocks and ranged from 0 to 52. 4. Dividends - the Annual Compound returns did not include dividends and are therefore understated (during the test period “Dividends averaged between 3% and 4% annually”) 5. Taxes, Commissions and slippage – “we assumed commissions of 2.5% on purchase price plus a 2.5% bid/ask spread (the bid/ask spread was applied to the 60% of our stocks that were purchased over-the-counter), a 2.5% commission on selling price, and a 25% capital gains tax (over 90% of the stocks were held long enough to qualify for capital gains treatment).” The commissions included are in line with the higher trading costs associated with the time period examined[v]. Furthermore, the inclusion of returns including taxes, commissions and slippage is highly instructive for a practitioner as it addresses concerns over “potential limits to arbitrage”. 6. P/E floating with bond yields – “We required a P/E corresponding to twice the prevailing triple A yield in each period (e.g. triple A yield = 8%; required PPE equal or below the reciprocal of 16%, or 6.25).” The Triple A bond yield during the period studied it ranged from approximately 7 to 9[vi] implying a requirement for a P/E below 5.5 to 7. This means that the P/E requirement of ≤ 5 for Portfolio 3 and 4 was less than the maximum floating P/E allowed for Portfolio 1 and 2. Having laid out the above we are better placed to analyze the results, bearing in mind the relatively small sample size, notwithstanding that the authors “attempted to select a statistically significant and unbiased sample of stocks”. The highest returning portfolio was Portfolio 4 (42.2%) which required stocks in the portfolio to possess the lowest valuation in terms of both liquidation value (≤ 0.85) and P/E ratio (≤ 5.0). Interestingly, Portfolio 3 (32.2%) returned more than Portfolio 2 (27.1%) despite its allowance for a higher liquidation cut off (≤ 1.0 vs. ≤ 0.85 for Portfolio 2) but a lower P/E threshold (≤ 5.0 vs ≤ ~5.5 to 7 for Portfolio 2). Portfolio 1 which allowed for the highest valuation in terms of liquidation value (≤ 1.0) and P/E (≤ ~5.5 to 7) generated the “lowest” return (20.0%). Overall, the results imply valuation drives futures returns, however, the merit of liquidation value vs the P/E ratio to explain future returns is less clear. In terms of relative returns all portfolios greatly outperformed the “OTC” and “Value Line” portfolios which generated an Annual Compound Return (before taxes and commissions) of 1.3% and -0.3% respectively. Significantly, the outperformance of all portfolios survived taxes, commissions and slippage. However, all returns during the period were severely impacted by the ravages of inflation which averaged approximately 7.5%[vii] per annum (geometric mean) from 1972 to 1976.



Conclusion


Despite the particularly short test period (6 years), limited portfolio holdings (average of 15) and the presentation of convoluted and seemingly irrelevant return data, “How the small investor can beat the market” still possesses utility. It demonstrated that the Annual Compound Returns for firms with positive earnings trading at or below liquidation value, combined with a low price to earnings ratio greatly outperformed OTC and Value Line firms during the examination period (April 1972 to April 1978) even after taking into account taxes, commissions and slippage. While the examination period was simply too short to provide definitive guidance as to the efficacy of investing in firms trading at or below liquidation value, the study nonetheless represents a small piece in the larger puzzle of determining what we “really know” about returns of such firms from an empirical standpoint.




Notes:

[i] MarketFox Interview with Rich Pzena in which he briefly discusses the study (https://i3-invest.com/podcasts/i3-podcast-marketfox-interview-with-rich-pzena/)


[ii] Approximately equal to an inflation adjusted USD 18m in 2020 (https://www.usinflationcalculator.com/)


[iii] https://en.wikipedia.org/wiki/Nasdaq


[iv] https://alphaarchitect.com/2015/04/01/where-to-find-cool-academic-finance-research/


[v] A Century Of Stock Market Liquidity And Trading Costs, Charles M. Jones, Graduate School of Business Columbia University, 2002 (https://papers.ssrn.com/sol3/papers.cfm?abstract_id=313681)


[vi] https://fred.stlouisfed.org/series/AAA


[vii] https://www.inflation.eu/inflation-rates/united-states/historic-inflation/cpi-inflation-united-states.aspx