The Number on the Page Is Not What You Think It Is
Reading Notes · The Intelligent Investor — Part 4
None of this is financial advice. I don't know your situation, your risk tolerance, or your time horizon. What follows is my attempt to distill what Graham actually says in the most technically demanding section of The Intelligent Investor — chapters 11 through 15 — because I think the analytical methodology here deserves more attention than the headline-grabbing quotes that get recycled everywhere.
I own no individual stocks discussed in these notes. My current portfolio is mostly index funds. I'm writing this series because reading Graham carefully has changed how I think about financial statements, not because I'm about to become an activist stock-picker.
Earnings Per Share Is a Story, Not a Fact
The single most important warning in chapter 12 is so obvious that most people skim past it: earnings per share is an accounting construct, not a measurement of economic reality.
Graham's concern, stated plainly: companies have enormous latitude in how they report earnings, and that latitude is routinely used in self-serving ways. He identifies three manipulation vectors worth understanding.
Special charges that aren't special. When a company reports "restructuring charges" or "impairment losses" as non-recurring items, it's implicitly asking you to ignore them in your valuation. The problem is that for many companies, these charges appear in the footnotes every single year, just under different labels. Graham's rule: if a "one-time" item recurs more than once in five years, it's not one-time. Include it.
Pro forma earnings versus GAAP earnings. Management can choose what to emphasize in press releases. "Adjusted" earnings that exclude stock-based compensation, amortization of acquired intangibles, or legal settlements are sometimes legitimate simplifications — and sometimes a way to report a number 30% higher than what the auditors signed off on. Graham would have recognized the pattern; Zweig's updated commentary in the 2003 edition names Enron and WorldCom as the inevitable consequence of taking management's preferred framing at face value.
EPS growth through share buybacks, not business growth. If a company earns $100M and has 100M shares outstanding, EPS is $1.00. If it buys back 20M shares and earns the same $100M next year, EPS is now $1.25 — a 25% "growth" with zero improvement in business performance. You need to check whether EPS growth is correlated with earnings growth, not just per-share arithmetic.
Graham's proposed solution is deceptively simple: use a multi-year average of earnings, not a single year. His suggestion is a 7–10 year average that smooths out both cyclical highs and special-item lows. This immediately makes a lot of exciting growth stories look more pedestrian — which is usually the point.
Security Analysis Without a CFA
Chapter 11 is titled "Security Analysis for the Lay Investor: General Approach," and it does something important: it distinguishes clearly between bond analysis and stock analysis, arguing they require different mental frameworks.
For bonds and preferred shares, the analysis is primarily defensive. The central question is coverage: does the company earn enough to comfortably service its obligations? Graham's general rule for industrial bonds requires earnings to cover interest charges at least 5x over the most recent 7-year period. For utilities, 4x. For railroads, 3x. These aren't arbitrary — they're calibrated to the volatility characteristics of each industry. A utility's revenue is predictable; a railroad's isn't. The coverage multiple accounts for that.
For common stocks, the framework shifts. Coverage matters less (common shareholders are residual claimants, not fixed-income holders), and earning power over time matters more. Graham argues that the primary value of security analysis for ordinary investors isn't to find hidden gems — it's to avoid mistakes. The analysis should tell you, with reasonable confidence, that a company won't collapse, its earnings are real, and the price you're paying isn't insane relative to what the business actually produces.
This framing — analysis as mistake-prevention rather than opportunity-finding — is worth sitting with. It reorients how you read a 10-K.
The margin of safety concept runs through both types of analysis, but differently. For bonds: safety margin is the cushion between coverage and minimum required coverage. For stocks: safety margin is the gap between estimated intrinsic value and market price. In both cases, you're buying protection against your own analytical errors. The margin isn't a prediction of upside; it's insurance against being wrong.
The Four-Company Comparison Method
Chapter 13 is one of the most pedagogically useful sections in the book, and it tends to get underappreciated because it's dense with numbers.
Graham takes four pairs of companies in related industries and performs parallel analysis. The method reveals something that's hard to articulate abstractly but obvious when you see it: quality differences between companies are visible in the financial statements if you know what to look at in combination.
The comparisons in the original 1973 edition are dated now, but the analytical structure survives. When you put two companies side by side — same industry, similar size — and look at:
- Operating profit margins (is the business competitively advantaged, or just riding an industry tailwind?)
- Return on equity over a full cycle (does the capital generate consistent returns, or only in boom years?)
- Long-term debt to equity (is the balance sheet conservative or stretched?)
- Price-to-book relative to industry peers (is the market pricing in quality, or ignoring it?)
...you start to see that "companies in the same industry" often aren't comparable at all. One has been compounding quietly for 15 years with stable margins; the other has higher revenue growth driven by acquisition and debt. The headline P/E might be identical.
Zweig's update points to Cisco versus Lucent in the late 1990s as a contemporary example of exactly this divergence becoming visible in retrospect. They looked similar from the top line. The balance sheets and cash flow statements told a different story.
The actionable lesson for today's investor: never analyze a stock in isolation. Find 2–3 peers and run the comparison. The contrast is informative in ways that looking at one company never is.
Seven Tests for the Defensive Investor
Chapter 14 is where Graham gets operationally specific. He proposes seven criteria for the defensive (passive) investor's stock selection. These are explicit filters, not vague principles:
1. Adequate enterprise size. Graham's original threshold was $100M in annual sales for industrial companies and $50M in assets for utilities. In today's dollars, you'd need to scale these substantially — Zweig suggests roughly $2 billion in revenue as a starting point for industrials. The point isn't the specific number; it's that very small companies carry higher operational risk and are often under-researched, which makes independent verification harder.
2. Sufficiently strong financial condition. For industrials: current assets should be at least twice current liabilities (a 2:1 current ratio), and long-term debt should not exceed net current assets. For utilities: debt shouldn't exceed twice the equity value at book. This is a conservative balance sheet test. Many technology companies would fail it entirely; that's intentional.
3. Earnings stability. No deficit (no net loss) in any of the past 10 years. This is a more stringent test than it appears. It eliminates cyclical companies that look fine most of the time but posted losses in 2009 or 2020. It also eliminates most early-stage growth companies by design.
4. Dividend record. Uninterrupted dividend payments for at least 20 years. Today this is more controversial, given how many high-quality businesses operate as non-dividend-paying compounders (Alphabet, Berkshire, Amazon). Graham was writing for a different era. The spirit of the criterion is: demonstrated history of returning capital to shareholders, in good times and bad. Whether that takes the form of dividends or buybacks is a legitimate question for contemporary application.
5. Earnings growth. At minimum, a one-third increase in EPS over the past 10 years, comparing 3-year averages at each end to smooth volatility. This is a real growth criterion — not demanding (roughly 3% per year), but it eliminates stagnant or declining businesses.
6. Moderate P/E ratio. Current price no more than 15x the average earnings of the past three years. At the time Graham revised the book, this was already somewhat conservative. In today's market environment, applying this strictly would eliminate most of the S&P 500. The principle — that you shouldn't pay extravagant multiples for earnings power — is sound even if the specific number requires contextual adjustment.
7. Moderate price-to-book ratio. Price no more than 1.5x book value. Combined with criterion 6, Graham suggests that the product of P/E × P/B should not exceed 22.5. So a stock trading at 9x earnings could trade at 2.5x book and still pass; a stock at 15x earnings can only trade at 1.5x book.
Run the Dow Jones 30 through these seven filters in any given year and you might find 3–8 companies that pass all of them. That's a feature, not a bug. The defensive investor isn't looking for opportunity; they're looking for safety. The goal is to own a basket of businesses that are unlikely to permanently impair your capital, not to maximize returns.
What the Enterprising Investor Does Differently
Chapter 15 is where the two-track system Graham has been building finally becomes explicit. The defensive investor minimizes analytical effort and accepts market returns from a filtered list. The enterprising investor does significantly more work in exchange for the possibility of materially better results.
The key difference isn't just higher P/E tolerance — it's the type of opportunity being pursued.
Second-tier companies. Graham argues that the market systematically undervalues companies that are "large but not prominent." Well-capitalized businesses in unglamorous industries, with solid earnings records, trading at 9–10x earnings because they don't attract institutional analyst coverage. The inefficiency isn't irrationality — it's neglect. The work for the enterprising investor is finding these situations before the market's attention catches up.
The net current asset value (NCAV) approach. This is Graham's most famous — and now most difficult to apply — method for enterprising investors. NCAV = current assets minus total liabilities (both current and long-term). The strategy: buy at two-thirds or less of NCAV. You're essentially buying the liquid assets of the business at a discount and getting the fixed assets and earnings power for free.
In Graham's era, this was a productive strategy. In the 1930s and 1940s, genuinely liquidation-value situations were common. Today, NCAV bargains are extremely rare in US markets; they occasionally appear in stressed sectors, micro-caps, or in Japanese and Korean markets where the value investing tradition has historically been undervalued. The principle is sound, but the hunting ground has narrowed.
The earnings yield plus stability screen. For situations that don't meet NCAV criteria, Graham proposes looking for companies with earnings yields (inverse of P/E) meaningfully higher than the going bond yield, combined with a strong balance sheet and relatively stable earnings history. The logic: if a stock has an earnings yield of 8% when Treasuries yield 4%, you're being compensated for the incremental risk of equity ownership. If that spread narrows to 1%, you probably aren't.
The practical challenge with all of these enterprising approaches today is that the research burden is real. Running a Graham-style NCAV screen in 1965 with paper financial statements was slow; today the data is cheap but the competition is intense. Every value screen I've seen discussed publicly gets arbitraged quickly once it becomes well-known.
What Holds Up and What Doesn't
I want to be honest about where Graham's specific numbers show their age.
The 15x P/E ceiling in criterion 6 would have kept you out of almost every profitable technology company for the last 25 years, including ones that compounded at 20%+ annually. The 1.5x book value ceiling is similarly restrictive in an economy where competitive advantage is increasingly intangible — brand, network effects, software — and doesn't show up on the balance sheet at all.
The author's conclusion in chapter 14 concedes this tension more than it's often given credit for. He says the defensive investor's criteria will eliminate "almost all" growth stocks, and that this is the point — the defensive investor shouldn't be in growth stocks because the analytical uncertainty is too high for a passive approach.
What holds up entirely:
The multi-year earnings average. This is just statistically correct. Single-year EPS is noisy.
The balance sheet conservatism. A company with 2:1 current ratio and manageable long-term debt is structurally safer than one financed to the hilt. This was true in 1949; it was true in 2008.
The skepticism about "special" charges. If anything, Zweig's commentary undersells how much more sophisticated earnings manipulation has become since 2003.
The comparative method. Looking at one company in isolation is always worse than looking at it in context. This is as true today as it ever was.
The overarching message of chapters 11–15 is one the book doesn't quite state this directly: most of what gets called "investment analysis" in financial media is pattern-matching on recent trends, not systematic evaluation of business quality and price. Graham's framework is tedious precisely because it's anti-intuitive — it asks you to slow down, use multi-year data, read footnotes, check the balance sheet, compare peers, and then ask whether the price you're paying leaves room for error.
The seven criteria don't guarantee you find a good company. They're designed to make sure you don't buy a bad one by accident.
That's a more modest ambition than most investment frameworks. It's also, I suspect, more achievable.
This is part 4 of a series on The Intelligent Investor. Part 1 covers chapters 1–3; part 2 covers chapters 4–7; part 3 covers chapters 8–10. Nothing in this series constitutes financial advice.
Leave a comment ✎