Security Analysis

The authors open by conceding the problem. Analysis is a method borrowed from science, and investment is not a science; skill and chance both decide outcomes, as they do in law and medicine. The book’s claim is narrower than its reputation: careful study of the facts improves the odds, and nothing more.

They sort the analyst’s work into three jobs: describing an issue, selecting among issues, and criticising them. Description — gathering the facts and setting them out plainly — is the part the manuals and statistical services already supplied. Selection, which means saying an issue is cheap, dear or sound, is where the analyst’s own judgment starts.

The chapter then retires two older ideas of what a stock is worth. Book value lost because prices and earnings showed no sign of tracking it. Earning power, stated as one firm number, lost because it assumes the past is a reliable guide, and experience kept showing that it was not.

What survives is looser. Intrinsic value, in their treatment, is a band rather than a point — a “range of approximate value” that widens as the facts grow less certain. They give Wright Aeronautical in 1922 a range of $20 to $40 and Case in 1933 a range of $30 to $130. A wide band still settles the question when the price sits far outside it. That is the whole use of the concept.

The authors split the evidence in two. The quantitative side is the statistical record: capitalisation, earnings and dividends, assets and liabilities, operating figures. The qualitative side is the nature of the business, its place in the industry, the quality of its managers and the outlook.

They lean on the numbers, for stated reasons. Figures are fewer, easier to obtain and better suited to firm conclusions, and the results already reflect much of what the qualitative factors would say. The qualitative answers come from scattered sources of uneven reliability, mixed with a good deal of opinion.

A caution runs the other way. The same kind of figure is worth different amounts in different trades. Five years of a railroad’s or a chain store’s earnings tell the analyst something; five years of a small oil producer’s may mislead, because price and output both change from year to year.

A straight bond, in this account, is a bargain with a capped upside: the holder gives up a share of profits and should get specific, convincing safety in return. From that the authors draw their sharpest line about method. In their words, “bond selection is primarily a negative art. It is a process of exclusion and rejection, rather than of search and acceptance.”

The asymmetry is the reason. A stock buyer who wrongly rejects a good issue pays for it in lost gain. A bond buyer can turn down any number of good bonds at almost no cost, provided the one finally accepted is sound. So there is no such thing as being too demanding.

Four principles follow. Safety rests on the issuer’s ability to meet all its obligations, not on a lien or other contract right. That ability should be judged in depression, not prosperity. A high coupon does not repair a weak issue — “Deficient safety cannot be compensated for by an abnormally high coupon rate.” And every bond should pass rules of exclusion and fixed tests of the kind the law set for savings banks.

The depression test gets its own chapter. “Any bond can do well when conditions are favorable; it is only under the acid test of depression that the advantages of strong over weak issues become manifest and vitally important.”

The authors refine the test. The real question is not whether an industry escapes depression but how much it swings. The steadier the business, the larger the share of its earnings that interest may take; the less steady, the wider the margin it needs. A company whose survival is in doubt in a long slump fails the qualitative test however large its past coverage.

Their summary looks back on the 1930s. Reasonably capitalised electric and gas companies came through, as did the few railroads with wide coverage before 1930. Among industrials even good records often failed, especially at small and middling companies, so they advise confining industrial bonds to firms of dominant size with a substantial margin of earnings over interest.

Convertible bonds, participating issues and bonds with warrants look ideal on paper: the safety of a bond and the upside of a stock. The authors note that the record has not matched the form, and give the first reason as simple. Few such issues ever met the standards of a sound investment; the privilege was usually offered to make up for weak security.

Their rule is that a privileged issue selling near or above face value must qualify either as a straight fixed-value investment or as a straight speculation in the common, and be bought with one of the two in mind. The investor should not relax his safety tests for a conversion right; the speculator should not take a mediocre company because a bond contract makes it feel safe.

Having knocked down both the old and the new-era methods, the authors try to rebuild. They argue that diversification can offset the instability of single companies, and that an earnings trend, dangerous as the only test, can still serve as one sign of merit among others.

The case for owning a diversified group of stocks bought at reasonable prices rests, in one version, on long-standing features of the American economy: growing national wealth, reflected in the profits of large companies, built mainly through new capital and retained earnings. That last link is what makes past records worth studying against current prices.

The authors frame the income statement as three questions: what were the true earnings, what does the record say about future earning power, and what standards should govern a valuation. They also set out Wall Street’s shortcut — take the last reported earnings per share and multiply by a factor for quality — as the habit the following chapters pick apart.

Outright false figures, they note, were rare among public companies; the Kreuger frauds of 1932 stood out for their boldness. The trouble lies in audited statements that are accurate within accounting’s limits and still need adjustment before an analyst can rely on them.

The chapter names “manufactured earnings”: large charges against surplus in a bad year that lower the cost base and so raise reported profit in the years after. The authors call this family of practices perhaps the most vicious kind of accounting manipulation — money taken out of surplus and later reported as income.

They add that the overwhelming majority of managements are honest, and in the next breath that loose or “purposive” accounting is “a highly contagious disease”. Chapter 34 carries the same charge to the write-down of plant: shrinking the asset base to cut depreciation and lift earnings is, in their view, an inexcusable subterfuge that auditors should refuse to certify.

After a case in which reported earnings for Tobacco Products common were overstated by about half, the authors state a rule of thumb: “where manipulation of accounts is found, stock juggling will be found also in some form or other.” The second half of the chapter turns to subsidiaries, whose results a parent’s own statement can understate or overstate depending on what it leaves out.

Using the past to judge the future is, the chapter says, “at once the most important and the least satisfactory aspect of security analysis.” Important because the past is the only clue; unsatisfactory because the clue is never wholly reliable and often worthless.

The authors set the average against the trend. A rising trend and a high average pull in opposite directions for the same current earnings, and the market gives the trend great weight. They warn that a trend may deceive, and that valuations built on it follow no arithmetic and so can be stretched without limit.

Here the book puts a number on the line between investment and speculation in stocks: “about 20 times average earnings is as high a price as can be paid in an investment purchase of a common stock.” A higher price is not called a mistake. It is called speculative, and a purchase above it that works out has proved a wise or lucky speculation.

Two corollaries follow. People who habitually pay more than about 20 times average earnings are likely to lose a good deal over time, because without a fixed check they give in to each bull market’s argument. And 20 times is a ceiling, not a target: for a company with neutral prospects the authors suggest about 12 or 12½ times average earnings, alongside a sound balance sheet and management.

The chapter sorts capital structures into overconservative, suitable and speculative. A speculative structure — heavy senior claims ahead of a thin common — takes every security of the company outside investment grade. In the authors’ worked example, a 25% rise in total earnings lifts earnings per common share by about half.

They note the other side: such companies tend to sell at high values in good markets and may be marked down further in bad ones, and when deflated their shares can rise much further than they can fall.

A price below liquidating value, the authors argue, should lead shareholders to ask their management hard questions. Managers usually answer that market prices are no concern of theirs. Graham and Dodd call that idea wrong and hypocritical: a fair price on sale is as important to an owner as earnings and assets.

Their sharpest line is about executives who claim not to watch the quote: “It is difficult not to lose patience with the sanctimonious attitude of many corporate executives who profess not even to know the market price of their securities.” Many, they add, have a personal stake in that price. They ask that directors be held to watch it and act on plain discrepancies.

Chapter 27 describes a market that set its standard of value by the price rather than the other way round. That is the question to put to any valuation that rests on an earnings trend running years ahead — the pattern the AI capex page tracks in the spending now committed. The book does not say such prices are wrong; it says they are speculative and should be called so.

Chapter 6’s first principle — safety comes from the issuer’s ability to pay, not from the lien — is the test for the borrowing behind the build-out. The collateral ladder dates the steps by which lenders were asked to take sites, leases and chips as security; Graham and Dodd’s point is that none of those pledges makes a weak borrower a sound one.

Chapters 32 to 34 are the guide for reading earnings that depend on depreciation choices, a live question for companies whose main assets are short-lived chips. The lens built on Graham is at Benjamin Graham, and the later, plainer version of these ideas is The Intelligent Investor.