More Than a Numbers Game: A Brief History of Accounting (Fintwit Book Club January 2025)

Summary

• Byrne Hobart’s core case for the book: accounting quality is a public good, and fraud’s blast radius extends to honest competitors. The episode’s best anecdote is WorldCom capitalizing line costs to look much more profitable than AT&T—which responded by firing 20,000 people and spending over $100B on cable companies, nearly destroying itself. Andrew’s takeaway: “I just never heard of a company almost getting destroyed by a competitor’s fraud before.”
• There is no good answer to who pays for the audit, and Super Micro is the live case study. Byrne walks the options—investors paying means duplicated work or a free-rider tax on the biggest holder—so companies pay, and investors learn to treat auditor identity as a signal. Andrew describes Super Micro’s Big Four resignation as effectively identifying the portions of the accounts not to trust. Byrne’s proposed successor play is a costly confirmation and massive restatement, with the hope of later moving back to a Big Four firm; Andrew thought a lower-tier mid-market firm took the engagement.
• A century of companies screaming that accounting changes would “destroy the capital markets”—and, according to the book’s studies as relayed by Andrew, markets almost never cared, on- or off-balance sheet, expensed or capitalized. Byrne’s stock-comp thought experiment says switching from half-stock to all-cash compensation while issuing enough stock to fund it changes nothing economic, apart from employee incentives and possible issuance/administrative-cost differences. “So if there’s any change in how you value a company… something is wrong with your accounting.” Companies that only look cheap ex-SBC, like Snap, have punished believers.
• Screen optics create short-to-medium-term mispricings that private equity may eventually arbitrage away. Andrew’s example: a company moved inventory financing on-balance sheet to save 50bps on hundreds of millions, worsening its screens for quants. Byrne counters that this shifts the shareholder base toward cash-flow investors, while PE—which “fixates on cash flow” and accepts persistent GAAP losses when the business is good—can eventually correct the mismatch. Andrew notes that this can take 3–5 years and an activist.
• Broken market-level rules of thumb are where the alpha is. Dow price-to-book sat around 1–2x from roughly 1920 to 1990, rose to 6x in the ’90s, and later fell back toward 4x; Andrew argues the Buffett indicator also looks different in a world of international firms. Byrne says P/E and price-to-sales-plus-growth may be breaking because “one company’s net dollar retention is another company’s lower steady-state gross margin” and AI businesses “may be a software business [but do] not have software margins”—some software may look “more like you’re investing in a steel mill than in Microsoft circa 1994.”
• Byrne’s non-hot-button candidate for the 2035 accounting debate: capitalizing more big-tech intangibles. He says Google’s true economic balance sheet feels “more like a 10% return-on-equity business,” once its algorithm, brand and culture are treated as accumulated capital. Andrew’s pushback is that market-value swings would make such accounting highly unstable; Byrne concedes he is exaggerating and says capitalizing more R&D and marketing is possible but “I don’t actually think it’s worth doing.” His tentative concrete complaint is SPAC-warrant mark-to-market treatment.
• Tax-code coevolution is underrated history: the book/Andrew point to 1981 accelerated depreciation as one Milken-era tailwind, while Byrne adds a Treasury-stripping basis-allocation loophole that let investors book immediate capital losses. Andrew calls it “basically an infinite money machine.” Byrne says Ronald Reagan’s cut in meal and entertainment deductibility from 100% to 50% helped destroy the Midtown dining scene; at a 92% top marginal rate, the three-martini lunch was “a 92% off happy hour,” and the tax code became “this massive cirrhosis subsidy.”

Deep dive

1. Why this book: accounting progress is invisible until you read the 1970s

• Byrne’s origin story: Supermoney (early ’70s) mentioned More Than a Numbers Game and marveled that some investors “literally go to the SEC and read this weird thing called a 10-K” rather than just the annual report—and even look at quarterly numbers rather than wait for annual earnings, “like saying some investors use limit orders.” Go back to 19th-century trading anecdotes and it’s reasonable to doubt participants knew what depreciation was. He wanted to know how accounting progress happened.
• His prior going in: “a lot of accounting judgments seemed superficially wrong until you try to come up with a better alternative.” His first remembered accounting change was the end of goodwill amortization, which Buffett endorsed—you would not depreciate the Coke brand to zero—yet cash flows did not change while investor-visible numbers did.
• The social stakes: reported numbers drive investor decisions, which drive competitor behavior—“good accounting is actually very, very socially useful because it’s not just a scorecard for any one firm.” Private businesses can account however they want; public companies have an obligation to present financial statements that accurately reflect reality.

2. Fraud’s blast radius: WorldCom nearly destroyed AT&T

• Andrew’s favorite anecdote: WorldCom capitalized line costs, looked much more profitable than AT&T, and made AT&T’s expensing look like a lazy dinosaur’s accounting. AT&T fired 20,000 people and bought over $100B of cable companies, nearly destroying itself. Andrew’s own practice—calling management to demand that it close a three-point margin gap to peers—suddenly looks dangerous when the peer is fraudulent.
• Byrne’s counter-texture via the Mechanical Turk: British inventor Edmund Cartwright fell for the fake chess robot but reasoned that a machine capable of playing chess implied a machine capable of weaving—and built an early power loom. “Sometimes the fraud does force people to step up their game.”
• On how frauds start: although some begin with an intent to lie, many arise from “just this one time, just this quarter, we are going to front-load just this one transaction.” Some 1990s frauds began by hiding profits. Enron’s trading desk was so profitable that it wanted a cookie jar for future write-offs and disliked the optics of California being in the dark while Enron was “milking them for all they’re worth.”

3. Arthur Andersen’s uniformity ideal, Enron’s rules-lawyers, and prop trading as the pro-social twin

• The through-line Andrew flags: Arthur Andersen is repeatedly described as the firm most committed to every accountant producing the exact same number—no discretion, everyone following the same rules. Andrew estimates that this desire and culture was probably “number six” among the hundred things that led to its demise.
• Byrne’s mechanism: accounting approximates rather than equals economic reality, so a sufficiently rules-driven auditor facing an Enron special-purpose entity—owned just enough not to consolidate, doing business with Enron and serving as collateral for derivatives on Enron’s own stock—could be “stuck saying that just per our ethics statement, we absolutely must approve this particular instance of lying to your shareholders.”
• Andrew’s reframe: Enron’s accountants in another life might have been elite prop traders, like the Polymarket traders who read the fine print after a government-shutdown deal and noticed that a technically hour-long shutdown could make the answer “yes.” Byrne agrees but sharpens it: prop trading can be “the pro-social application of the same skill”—an adversarial “distributed bug-bounty program” that pressures mispriced structures and helps keep markets efficient. The same skill in a cooperative setting is much harder to police.
• Byrne extends the incentive problem to audits. If investors pay, there is either one audit per investor and duplicated work, or one investor pays while everyone else benefits. The least-bad answer may be for the company to pay, with investors learning which auditors are a positive signal, neutral, or a warning.

4. Who pays for the audit? No good answers—and Super Micro proves it

• The structural dilemma as Byrne relays it: investor-paid audits mean either duplicated work per investor or the biggest holder “paying a tax to keep the rest of the market informed.” Andrew jokes that the top two holders could sell shares back and forth to pass the bill around.
• The industry cannot easily sustain a stack-ranked reputation among the large firms—“it’s just really hard to be the second most reputable of the big firms.” The practical tiers are Big Four versus smaller auditors, where hiring a small firm may mean either prudent cost control or “you can’t afford to have someone from EY looking at your books.”
• Andrew’s live example: Super Micro rose roughly 5x from January to March—his estimate was that it contributed about 1.5 percentage points to a 10% Russell 2000 year—before graduating to the S&P 500. It then faced accounting issues, insider-trading concerns and a short report, and the stock fell 80%.
• Super Micro’s Big Four auditor resigned while the company was still a roughly $50B enterprise-value business. Andrew characterized the resignation as the kind of letter that, point by point, tells readers which parts not to trust. He thought one lower-rung mid-market firm took the engagement.
• Byrne’s proposed playbook is to hire the new auditor to confirm what the previous auditor said, conduct “a massive restatement,” and pay heavily for the work. The company would hope to move back to a Big Four firm later; meanwhile, the mid-market auditor could tell prospective clients that it audits an S&P 500 member.

5. Modern accounting is two merged schools that never agreed on the question

• Byrne’s historical frame for why accounting is “annoyingly complicated”: school one is the London bond investor lending to “the hot tech stocks of the day, which are American railroads,” caring whether $1M invested is backed by $1M of assets. School two is manufacturing cost accounting—General Motors product mix, assembly lines, component costs and whether to make one model rather than another. One focuses on downside protection and liquidatable collateral; the other focuses on flows, costs, profit and upside.
• Those are different questions—one about a stock of physical assets, the other about flows and their interrelation—but balance sheets must link to P&L and cash-flow statements, so the schools had to merge despite different assumptions.
• The best illustration of legitimate judgment calls is the investment tax credit. Buy a $1M machine and receive a $60,000 credit: did you actually spend $940,000 and depreciate that amount, or did you receive $60,000 of revenue in response to a business action? “You can kind of see it both ways.”
• Accountants often choose the more cautious of two equally defensible treatments because “you never want to give someone an incentive to do something more aggressive” or reward maximum risk-taking similarly to prudent decisions.

6. Every decade, companies screamed “this will destroy us”—and markets often did not care

• Andrew’s overall takeaway: chapter after chapter, companies insist that bringing off-balance-sheet debt on-balance sheet, expensing stock compensation or making another disclosure change will “destroy the capital markets.” The author then dryly cites studies suggesting that markets did not care much about off-balance-sheet versus on-balance-sheet treatment or expensing versus capitalization. Andrew suggests that EPS and share-count treatment in the 1990s may be a partial exception.
• Byrne’s twist: that reaction can actually be to the companies’ credit. For a change not to matter, they must have been behaving fairly economically rationally; “if they were gaming it, then it absolutely would destroy them.”
• Stock comp is the exhibit. Companies that look cheap only if SBC is ignored, such as Snap, have not rewarded investors who relied on that valuation. Yet Andrew relays the companies’ side: tech companies such as Snap and Twitter have asked what they are supposed to do when competing with Facebook and Google for engineers while spending 8% of sales on stock compensation.
• Byrne notes that Meta and Zoom have moved more toward cash compensation as they can afford it and want clearer economics. His thought experiment is that switching from half cash/half stock compensation to all-cash compensation while issuing enough stock to fund the cash should not change the company’s economics—except insofar as employees respond differently to the incentives, or issuance and underwriting costs differ from the administrative cost of granting options. If valuation changes despite those qualifications, “something is wrong with your accounting.”
• Andrew’s rejoinder is, “Do you like accounting or do you like making money?” Byrne replies that he has never heard someone say that during a bear market.

7. Screens versus economics: quant optics, cost of capital and private equity

• Andrew’s live dilemma: a company moved inventory financing on-balance sheet, adding debt that peers keep off-balance sheet, to save 50bps on hundreds of millions. Economically the decision is better, but it screens worse “in a world of passive” investing.
• Byrne’s answer: quant strategies may be long 600 stocks and short 800, or similarly diversified, so the pressure from one changed signal is incremental even when many strategies use similar signals. The company may be engineering turnover in its shareholder base—fewer quants and indexers, more investors focused on cash flow and economics.
• On Andrew’s question about concentrated investors having a higher cost of capital, Byrne says quants think about capital differently. A diversified strategy focuses on incremental volatility, beta, required equity and collateral for the prime broker rather than simply the broker’s financing charge. Quants may estimate opportunity cost more accurately and reasonably target a higher return on the equity slice of their strategies.
• Byrne’s own current position illustrates screen-blindness: a high-margin Polish manufacturer bought a distributor that added a large amount of revenue and little profit. It now screens as an average-margin industrial company and is also being kicked out of an index. “If a rogue asteroid destroyed the distribution company, I think the stock would probably go up.”
• The ultimate long-run corrector is private equity, which “fixates on cash flow,” models balance sheets, P&Ls and cash flows, and is willing to own businesses whose S-1s show persistent GAAP losses despite a viable cash-generating business. Andrew notes that PE generally does not go hostile, so an inefficient company might persist for 3–5 years until an activist forces a sale. He also observes that with diversified long-term bets, “it’s always somebody’s short term.”

8. Milken’s tailwinds, infinite money machines and Reagan killing the three-martini lunch

• A related structural mispricing Andrew highlights: by-rating Sharpe ratios suggest that buying the highest-rated junk bonds has produced the highest Sharpe, while buying CCC-rated bonds just before default has produced the lowest, partly because of lottery-ticket demand. BBB bonds are also unattractive because issuers can optimize to be as levered as possible while retaining investment-grade status, and there is “infinite appetite for that particular kind of paper.”
• Andrew hypothesizes that a leveraged long-BB/short-BBB trade might be attractive, but funding, liquidity and sizing could prevent it. The BB universe may be only one-tenth the size of the BBB universe.
• The book’s Milken discussion, as Andrew relays it, points to tax-code tailwinds including the 1981 allowance for accelerated depreciation. Byrne adds from rereading Predator’s Ball that Milken’s brother was skilled with taxes. Under the then-existing Treasury-stripping rules, an investor could buy a Treasury, sell off different components and allocate basis between them as desired. Selling a zero-coupon piece could create an immediate capital loss while Treasuries yielded roughly 15%, allowing very rapid after-tax compounding.
• Andrew calls this “basically an infinite money machine.” On why such loopholes existed, Byrne cites information scarcity: “Control-F is just a wonderful technology,” and if he could bring one tool back to 1955, it would be Control-F plus a digital tax code—though he would probably end up in prison.
• Byrne’s other explanation is coevolution between the tax code and behavior. Ronald Reagan’s reduction of meal and entertainment deductibility from 100% to 50% helped destroy the Midtown dining scene. At a 92% top marginal rate, business drinks were “a 92% off happy hour,” helping create the three-martini lunch and a business culture around it. Byrne jokes that the tax code became “this massive cirrhosis subsidy,” making people dysfunctional from 1 p.m. onward.

9. Hindsight bias, the case for nominal accounting and the Macy’s trap

• Andrew’s question: the book makes crashes look foreseeable. Its account of loan-modification treatment in the 1970s—keeping troubled loans at par if repayment was still expected—was argued to have helped set up the savings-and-loan crisis, and Andrew compares it with later held-to-maturity and mark-to-market problems. He asks whether historical parallels help in real time or whether hindsight calls “10 of the last one crashes.”
• Byrne says there will not be a carbon copy of Enron, but there will be companies whose net income looks good while cash flows do not correspond to it, or that follow the letter of the law while making the business look better than it is.
• The inflation chapter changed Byrne’s view of nominal accounting. In the book’s truck example, replacement costs rise so quickly that economic depreciation makes the trucking business “a net destroyer of capital.” But adjusting for inflation invites endless debates over which exact truck or CPI component is relevant, while many company obligations are nominal. His division of labor is for analysts to ask replacement-capex and cash-flow questions while accountants choose “an answer that is 85% right but everyone can understand the logic, versus 99% right and we can spend forever debating.”
• Andrew’s counter-case is department stores, which he calls the number-one destroyer of value investors’ capital over the past 15 years. He relays the common Macy’s argument: roughly $5B of market capitalization against real estate valued at $8–14B, plus roughly $400M of reported profit. If $10B of real estate could generate $800M independently, the retail operation would be destroying roughly $400M of annual value, while management does not appear eager to liquidate the business and eliminate its own jobs.
• Andrew and Byrne discuss why the gap might persist. Andrew speculates that Amazon might pay roughly twice the current share price for the real estate, but political pressure, the Macy’s name, and the risk of selling piecemeal could prevent any buyer from extracting the value. That may leave the stock at an equilibrium between the real estate value and the negative value of the retail cash flows. Andrew makes a similar point about U.S. Steel, saying a differently named company might have been able to complete the Nippon deal.
• Andrew also notes that the Macy’s Thanksgiving Day Parade may be receiving about $200M for a roughly 20-year broadcast arrangement with NBC and Peacock—figures he is not certain about—and estimates that at around $2B of NPV against a roughly $4.5B company value. He is unsure whether Macy’s receives the money or who owns the parade rights.
• Andrew argues that Macy’s brand name may itself be a liability: the company cannot simply mark up brand equity when the brand may make a sale politically harder. He adds that the logo’s star was reportedly based on founder Mr. Macy’s sailor tattoo.

10. The 2035 accounting debate and today’s breaking rules of thumb

• Byrne’s honest non-answer on the next decade’s fight: the real issue “doesn’t actually qualify for hot-button treatment.” He thinks big tech should capitalize more intangibles because Google’s true economic balance sheet feels “more like a 10% return-on-equity business,” with the algorithm, brand, employee cohesion and culture treated as accumulated capital.
• Andrew’s pushback is that Google’s market capitalization went from roughly $400B two years earlier to roughly $1T when they spoke. If intangibles were marked to market, a 2022-style 30% decline in large-cap stocks would create an unstable accounting exercise. “It’s obviously a shortcoming of accounting, but I feel like it’s settled math.”
• Byrne says he is exaggerating. One possible approach would be to capitalize more R&D and marketing, as investors already implicitly do for some SaaS companies by capitalizing sales costs over customer-contract lives. But after the disruption caused by even a moderate SaaS-accounting change, “I don’t actually think it’s worth doing.” His tentative concrete complaint is that SPAC warrants should not have to be marked to market; he calls that the SEC’s next-best response after being unable to ban SPACs.
• Byrne’s related thesis is that capital intensity is a feature of where a company is in its cycle, not necessarily of its industry. In the 1990s debate over Amazon and eBay, eBay traded at a premium for being asset-light, until Amazon’s heavier investment won on shipping speed and selection.
• The breaking market-level indicators, in Andrew’s framing, include the old Dow price-to-book range of roughly 1–2x, its rise to 6x in the 1990s and subsequent decline toward 4x, as well as the Buffett indicator’s changed meaning when many firms are international. Byrne says P/E and price-to-sales-plus-growth assumptions may both be breaking. High-quality recurring revenue makes this year’s P/E less informative, while “one company’s net dollar retention is another company’s lower steady-state gross margin.” AI “may be a software business [but] does not have software margins,” because every interaction has a substantial incremental cost. Some software businesses may look “more like you’re investing in a steel mill than in Microsoft circa 1994.”
• Andrew’s instinct that an acquirer can always rip out costs meets Byrne’s integration-moat argument. Companies still expect leverage from employee and other operating costs, but a web of integrations—such as a Slack bot that prepares a Zoom meeting from Dropbox files—can make switching providers “a giant technical lift.” That inconvenience does not appear as an asset on Zoom’s balance sheet but is “absolutely a source of incremental DCF dollars.”
• Andrew closes with Lotus Notes: the book’s author worked at IBM with Lotus Notes, and some large firms may still use the difficult-to-replace system even today because replacing it would destroy the integrations built around it.