Buffettology · Chapter V

Financial *Metrics* That Matter

GAAP financial statements are the starting point, not the destination. This chapter covers the metrics Buffett actually uses to see through accounting to the underlying economics — Owner Earnings, ROIC, cash conversion, balance sheet safety, and the original 1989 concept of Look-Through Earnings.

Level:Advanced
◆
Read time:30 min
◆
Chapter:5 of 8
The Buffett invention
Owner Earnings
1986 letter
The honest return
ROIC > 15%
Sustained over a decade
The cash test
OCF / Net Income
≥ 90% sustained
The 1990 example
$590M look-through
vs $371M GAAP
Why this chapter exists

See through accounting to economic reality

GAAP exists to standardize what gets reported. It does not exist to reflect what a business is actually worth or what it is actually earning. The two diverge in subtle and predictable ways, and learning to see the divergence is most of the work in financial analysis. Buffett's repeated framing across decades: financial statements are the starting point, never the destination.

This chapter is structured around three acts. The first covers Owner Earnings — Buffett's 1986 reformulation of "earnings" that corrects two of GAAP's largest distortions. The second covers the four pillars of metric analysis Buffett uses for screening: return on capital, cash quality, balance sheet safety, and earnings-quality red flags. The third covers Look-Through Earnings — Buffett's 1989 invention that fixes GAAP's blindspot for non-controlling equity stakes.

“
Key takeaway
One sentence summary: every metric in this chapter exists to answer a single question Buffett asks of every business — "how much does this company truly earn?". Net income will not tell you. EBITDA will mislead you. The metrics here will triangulate the answer from multiple angles.
Act I
I

Owner Earnings

Buffett's 1986 reformulation of 'earnings' — the closest thing to a true economic income statement that GAAP allows you to construct.
The 1986 definition

Owner Earnings — what a business actually generates for its owners

Buffett introduced this concept in the 1986 shareholder letter as the best available tool for measuring a business's true earning power. The reasoning: net income overstates economic earnings (because it ignores ongoing maintenance capex needs) and EBITDA overstates them more (because it pretends assets do not wear out at all). Owner Earnings sits between them and is closer to what a long-term owner actually has available to spend.

Owner Earnings — Buffett's 1986 formula
Owner Earnings = Net Income + D&A − Maintenance CapEx − Required increase in working capital
D&A is added back because it is a non-cash charge against accounting earnings. Maintenance capex is subtracted because it is a real cash outlay required to keep the business at its current competitive position. Working capital increases tie up cash even when not appearing as 'expense' on the income statement.
What 'maintenance' means here is exact, not approximate
Buffett's 1986 phrasing is deliberate: maintenance capex is the cash needed to maintain competitive position, not just physical operations. A retailer that does not refresh its store footprint will see foot traffic decline. A software company that does not invest in product development will see product obsolescence. The cash needed to keep market share, brand relevance, and competitive parity is all maintenance — even when it does not look like physical asset replacement on the books.
01
Worked example
A worked example — capital-light business

Imagine a software company reporting the following annual numbers (simplified, in millions):

  • Net Income: $500
  • Depreciation & Amortization: $80
  • Total Capital Expenditure: $90
  • Working Capital Increase: $15

Management discloses (in the MD&A) that approximately $70 million of capex was for maintenance — keeping data centers refreshed, product platforms current, security infrastructure updated. The remaining $20 million was for new product launches. Owner Earnings calculation:

$500 (Net Income) + $80 (D&A) − $70 (Maint. capex) − $15 (WC) = $495 million

The number sits roughly equal to net income — typical for an asset-light business where depreciation closely matches maintenance needs. FCF (operating cash flow minus total capex) would be lower at around $475 million because it deducts the growth capex that is actually creating future value rather than maintaining current capacity.

02
Worked example
A worked example — capital-heavy business

Now an industrial company with similar reported earnings but very different underlying economics:

  • Net Income: $500
  • Depreciation & Amortization: $300
  • Total Capital Expenditure: $420
  • Working Capital Increase: $30

Industry analysis suggests maintenance capex runs roughly 1.2× depreciation for this type of asset-heavy business — replacing equipment costs more than the historical depreciation reflects because of inflation in capital goods prices. Maintenance estimate: $300 × 1.2 = $360 million. Owner Earnings:

$500 (Net Income) + $300 (D&A) − $360 (Maint. capex est.) − $30 (WC) = $410 million

Owner Earnings sits 18% below reported net income. The two businesses look the same on the income statement but are fundamentally different economic propositions. The first throws off cash for shareholders. The second consumes most of its earnings just maintaining the asset base.

The estimation problem is real but tractable
The hardest step is splitting total capex between maintenance and growth. Three practical approaches: read the MD&A and earnings calls for explicit disclosure (some managements provide it; most do not); use depreciation as a starting estimate (multiplied by an industry-specific factor — close to 1.0× for asset-light, up to 1.5× for capital-intensive industries facing rising replacement costs); compare against industry peers' multi-year capex/depreciation ratios. The goal is not precision; it is avoiding the much larger error of treating reported net income as economic income.
The capex split

Maintenance versus growth capital expenditure

The single most important distinction in Owner Earnings analysis is between maintenance capex (real economic cost, must be subtracted) and growth capex (optional investment in future returns, does not reduce current earnings power). GAAP combines them into one line on the cash flow statement. The investor must split them.

Dimension
Maintenance CapEx
Growth CapEx
What it is
Cash needed to keep the business at its current competitive position
Cash deployed to expand into new capacity or geography
Mandatory or optional?
Mandatory — skip it and the business deteriorates
Optional — management's allocation choice
Treatment in Owner Earnings
Subtract from earnings (real economic cost)
Do not subtract (this is value creation, not maintenance)
Why most reports lump them together
GAAP does not distinguish; both appear as 'capex' in cash flow statement
Same — investor must estimate the split
Practical estimate
Often close to depreciation in mature businesses
Total capex minus maintenance estimate
Industry archetypeTypical maintenance/depreciation ratioReasoning
Software, brands, services0.7×–1.0×Asset-light; depreciation tends to overstate replacement need
Consumer products, light manufacturing1.0×–1.2×Roughly steady-state replacement
Industrials, machinery1.1×–1.4×Equipment replacement costs rise with inflation
Heavy capex (utilities, telecom, oil)1.2×–1.6×Long-cycle assets, replacement costlier than original
Cyclical commodity (steel, mining)Varies widelyMaintenance can collapse to zero in downturns
◆Analyst note
One useful sanity check: for any business, the cumulative maintenance capex over a multi-year period should be roughly consistent with what is needed to keep capacity stable. If estimated maintenance capex is well below depreciation for many years in a row, you are probably under-estimating it — the business is implicitly running down the asset base. If maintenance capex is consistently above 1.5× depreciation, inflation in capital goods or accelerating obsolescence is likely the cause; both are real economic costs that owner earnings must reflect.
Why EBITDA fails this test entirely
EBITDA adds back all depreciation without subtracting any maintenance capex. The implicit claim is that assets do not need to be replaced — which is false for every business that owns physical assets. The discrepancy between EBITDA and Owner Earnings is exactly the cash that must be spent to keep the business running but that EBITDA pretends is free. For a steel mill, EBITDA can overstate distributable cash by 50%–80%. For a software company, the gap is smaller. The metric is not uniformly wrong; it is structurally biased toward businesses that need heavy reinvestment, which is the opposite of the right bias.
Dimension
EBITDA
Owner Earnings
What gets added back
All depreciation and amortization
Depreciation only — but maintenance capex is then subtracted separately
Treats assets as wearing out?
No — implicitly assumes assets are eternal
Yes — recognizes real cash needed to maintain capacity
Useful for capital-light businesses?
Distorts upward — adds back D&A that isn't economically real
Roughly equal to FCF when capex ≈ maintenance
Useful for capital-heavy businesses?
Severely misleading — flatters the economics
Reveals the true cash drag of asset replacement
Used by leveraged buyout shops?
Heavily — because it inflates apparent cash flow
Rarely — because it tells the truth
Buffett's view
Called "EBITDA financial fiction" in multiple letters
Defined this concept in 1986 letter as the right alternative
Act II
II

The four metric pillars

Return on capital, cash quality, balance sheet safety, earnings-quality red flags. The screening framework Buffett uses to size up an unfamiliar business in twenty minutes.
Pillar one

ROIC — return on invested capital

Return on Invested Capital (ROIC) measures how much operating profit a business earns per dollar of capital employed in the business — equity, debt, and minority interests combined. It is the cleanest cross-comparison metric in financial analysis because it strips out the financing decisions and reveals the underlying operating economics.

ROIC formula
ROIC = NOPAT ÷ (Equity + Interest-bearing Debt − Excess Cash)
NOPAT (Net Operating Profit After Tax) = Operating Income × (1 − tax rate). The denominator captures the total capital employed in operations, excluding cash that isn't being used. Two companies in the same industry with very different debt levels can have the same ROIC if their underlying operations are identical — which is exactly what makes it useful.
“
Key takeaway
ROIC is the single most important financial metric in Buffett's framework, mentioned by name or by concept in dozens of shareholder letters. The reason: a business earning ROIC above its cost of capital creates value with every dollar reinvested. A business earning ROIC below its cost of capital destroys value with every dollar reinvested. The threshold is binary, and the implications compound for decades.

ROIC > 20%

Exceptional. Strong moat in operation. Each retained dollar reinvested in the business creates approximately twenty cents of value annually. Sustained at this level for a decade-plus, the business is producing the kind of long-term compounding that defines Buffett's largest holdings.

ROIC 15%–20%

Excellent. Buffett's typical screening threshold. Genuine competitive advantage, with real value creation on every reinvested dollar. The bulk of Berkshire's equity portfolio holdings have traded in this range over their ownership periods.

ROIC 10%–15%

Acceptable. Above cost of capital for most companies. Value is being created but at a modest pace. Sustainability matters more than the level — a business defending 12% ROIC over twenty years can be a more reliable compounder than one that occasionally hits 25%.

ROIC 6%–10%

Marginal. Roughly at or below cost of capital for most businesses. Reinvestment creates little or no value. A growing company in this range is destroying capital quietly while looking productive.

ROIC < 6%

Value destroying. Capital is being systematically destroyed. The longer the business operates at this level, the more shareholder wealth is consumed. Many commodity industries (airlines, generic textiles) have decade-average ROICs at or below this level.

The trend matters most

A business with declining ROIC is showing moat erosion in real time, regardless of the absolute level. A business with improving ROIC is widening its moat, even if the current level looks unimpressive. 10-year ROIC trend is often more diagnostic than the current quarter's number.
03
Worked example
The reinvestment math, made concrete

A company retains $1 billion of earnings instead of paying dividends. Three different ROICs produce three very different outcomes after ten years of reinvestment at that rate:

  • ROIC of 25%: $1B compounds to roughly $9.3B in incremental value over a decade.
  • ROIC of 12%: $1B compounds to roughly $3.1B in incremental value.
  • ROIC of 6%: $1B compounds to roughly $1.8B — but if cost of capital is 8%, this represents destruction of value relative to what shareholders could have earned by receiving the cash.

Same $1 billion of retained earnings; three radically different outcomes for shareholders. ROIC determines whether retained earnings build wealth, hold steady, or erode it. This is why Buffett's letters return to ROIC so consistently — it is the metric that connects business quality to compounding outcome.

Related but different

ROE — and the leverage trap

Return on Equity (ROE) measures net income as a percentage of shareholders' equity. It is the more familiar metric, the one most investors learn first, and it has a serious structural problem: it can be artificially inflated by leverage without any improvement in the underlying business.

ROE formula
ROE = Net Income ÷ Average Shareholders' Equity
Buffett's threshold: 10-year average above 15%, with explicit verification that the level is not the result of excessive leverage. The verification matters — without it, ROE flatters levered businesses and obscures the underlying operating quality.
Dimension
ROE
ROIC
What's in the numerator
Net Income (after interest expense and tax)
NOPAT — operating profit after tax, before interest
What's in the denominator
Shareholders' equity only
Equity + interest-bearing debt − cash
Affected by leverage?
Yes — debt inflates ROE without improving operations
No — measures returns on the entire capital base
Comparable across capital structures?
Poor — two identical operating businesses can show very different ROE
Strong — strips out financing decisions
Buffett's standard
10-year average above 15%, with explicit check on leverage
Sustained above 15%; above 20% is exceptional
When to use which
First-pass screen; useful inside an industry
Cross-industry comparison; capital allocation decisions
The DuPont decomposition reveals the leverage component
Three drivers of ROE: net margin (income/revenue), asset turnover (revenue/assets), and equity multiplier (assets/equity). The first two reflect operations. The third is leverage. A 20% ROE from 5% margin × 1.5× turnover × 2.7× leverage looks identical to a 20% ROE from 12% margin × 1.4× turnover × 1.2× leverage — until the cycle turns and the heavily levered business cannot cover its interest payments. ROIC strips this distortion out, which is why it ranks higher in Buffett's framework.
Pillar two

Cash Conversion Ratio — earnings quality in one number

Reported earnings include accruals — revenue recognized before cash is collected, expenses recognized before cash is paid. Over short periods, accruals can diverge meaningfully from cash. Over long periods, they should converge. The Cash Conversion Ratio measures how reliably this convergence happens.

Cash Conversion Ratio
Cash Conversion Ratio = Operating Cash Flow ÷ Net Income
Sustained above 90% indicates high earnings quality — most reported earnings show up as cash. Persistently below 70% requires investigation: either receivables are bloating, inventory is building, or revenue recognition is aggressive relative to cash collection. One bad year is not diagnostic; three consecutive years below 70% almost always reveals a structural problem.
The five-year cumulative test
For a clean diagnostic, take five years of cumulative net income and compare to five years of cumulative operating cash flow. Healthy businesses show the two roughly equal. When five- year cumulative cash flow is materially below five-year cumulative net income, accruals are growing structurally — and structural accrual growth is, over time, accounting fiction being recorded as economic reality. This single test surfaces most major earnings-quality failures.
04
Worked example
The pattern that preceded major accounting failures

Most large accounting fraud cases of the last 25 years showed the same statistical pattern in the years before they were uncovered: high reported earnings, materially lower operating cash flow, growing receivables, growing inventory, increasing use of "adjusted" metrics. The pattern was visible in financial statements long before the disclosures came. WorldCom showed it. Tyco showed it. Lehman's commercial real estate loans showed it. Wirecard showed it spectacularly. The auditors saw it in some cases and did nothing; in other cases the auditors flagged it and were ignored.

The defensive lesson is not that every divergence between earnings and cash signals fraud — most divergences are benign and explainable. The defensive lesson is that the metric is cheap to check and high signal when wrong. Spending five minutes calculating five-year cumulative cash conversion is one of the highest-leverage diagnostic tests available.

Pillar three

Balance sheet safety — surviving the bad years

Buffett has often said that what kills companies is not bad results — it is bad results combined with leverage that does not allow time for recovery. A business that earns 6% on capital forever is mediocre. A business that earns 6% on capital with no debt remains mediocre and survives. The same business with heavy debt and short-term maturities can fail entirely during a downturn, taking shareholders to zero on what looked like recoverable economics.

Debt / EBITDA < 2.0×

Comfortable range. Even sustained earnings declines should leave room to service debt. Most of Berkshire's equity holdings over the decades have operated meaningfully below this threshold.

Debt / EBITDA 2.0×–3.0×

Watchable. Manageable in normal economic conditions but vulnerable during cyclical downturns. Capital-intensive industries (utilities, telecom) routinely operate here, with the safety provided by stable cash flows rather than balance sheet conservatism.

Debt / EBITDA > 3.0×

Dangerous, especially with cyclical or interest-rate-sensitive earnings. Buffett's typical reaction: pass. Even good businesses become very different propositions at this leverage level — a 30% revenue decline can convert a viable company into a bankruptcy candidate.

Interest coverage > 5×

EBIT covers interest expense by 5× or more — strong cushion. For investment-grade analysis, interest coverage is often treated as more diagnostic than debt-to-EBITDA because it measures the actual ability to service debt from operations.
The 30% revenue decline test
A practical thought experiment Buffett has framed across multiple letters: assume the business's revenue falls 30% next year. Can it still repay maturing debt? Can it sustain core operations without dilutive financing? If the answer to either is no, the position size needs to be smaller than the underlying business quality would otherwise suggest. The 2008 cycle produced multiple cases of fundamentally good businesses destroyed not by economics but by debt structures that worked in good times and broke in bad ones. Pre-mortem the balance sheet before adding the position, not after.
Pillar four

Earnings quality red flags — the eight signals

Buffett's letters and writings across decades produce a consistent list of warning signs that a company's reported earnings may not reflect underlying economic reality. None of them are individually disqualifying. Two or more appearing together warrant serious investigation. Three or more essentially demand it.

SignalWhy it matters
Receivables growing faster than revenueChannel stuffing or aggressive revenue recognition
Inventory growing faster than revenueDemand softening, future writedowns likely
Operating cash flow persistently below net incomeEarnings exist on paper, not in cash
Heavy use of "adjusted" earningsGAAP being filtered to look better than reality
Goodwill balance from premium acquisitions + mediocre ROICCapitalized adrenaline, not economic goodwill
Frequent non-transparent related-party transactionsPossible self-dealing
Auditor change or qualified audit opinionMaterial disagreement was likely involved
Insiders selling material amountsPeople with most information are reducing exposure
◆Analyst note
The red flags are useful as a screen, not as a verdict. Many companies show one or two of them at any given time for benign reasons — a single quarter of receivables outpacing revenue because of a large enterprise deal closed late, an auditor change driven by routine rotation rules. The diagnostic value comes from clusters, persistence, and the absence of clear innocent explanations. Run the screen across three years of financials before any new investment.
The integrity-of-management connection
Several of these red flags overlap directly with the integrity signals discussed in Chapter IV. Heavy use of adjusted earnings, non-transparent related-party transactions, auditor changes, insider selling — these are simultaneously financial signals and character signals. A clean financial profile from a CEO with warning signs on integrity is more dangerous than a messier profile from a transparently honest one. The numbers and the people are not separate analyses.
Act III
III

Look-Through Earnings

Buffett's 1989 invention to fix GAAP's blindspot for non-controlling equity stakes — and why it matters whenever a company holds substantial minority investments.
An original Buffett concept

Why GAAP underreports a holder of equity stakes

GAAP has a structural problem with equity ownership below 20%. When a company holds a small stake in another business, GAAP allows recognition of dividends received but not of the proportionate share of underlying earnings. For high-quality businesses retaining earnings to compound rather than distributing them, this means the holder's books show almost nothing while their economic share of value creation is substantial.

Buffett introduced Look-Through Earnings in the 1989 shareholder letter to correct this distortion when analyzing Berkshire. He calculated and disclosed the metric publicly for eight consecutive years (1990–1997) before suspending it after the General Re acquisition changed Berkshire's structure. The disclosure has continued in modified form since 2018 — comparing dividends received from investees against their proportionate retained earnings — but the underlying concept remains the cleanest way to think about earnings power for any company holding substantial minority equity stakes.

Look-Through Earnings — the 1989 formula
LTE = Operating Earnings + (Investee Retained Earnings × Ownership %) − Hypothetical Tax on Distribution
The hypothetical tax adjustment recognizes that retained earnings would face additional tax if actually distributed — typically a small adjustment relative to the proportionate earnings being added. Buffett's 1990 calculation: $250M proportionate retained earnings, $30M hypothetical tax, net addition of $220M to reported operating earnings.
05
Worked example
The 1990 Berkshire calculation

Buffett's own worked example, from the 1990 shareholder letter:

  • Berkshire's reported operating earnings (excluding capital gains and non-recurring items): $371 million
  • Proportionate share of investee retained earnings: $250 million
  • Hypothetical tax if those retained earnings had been paid as dividends: $30 million
  • Look-Through Earnings: $371M + $250M − $30M = $590 million

The difference between the GAAP $371M and the Look-Through $590M represents $219 million of earnings being created on Berkshire's behalf each year that GAAP simply did not capture. Cap Cities/ABC alone contributed over $82 million of proportionate earnings while only about $530,000 appeared on Berkshire's books (the dividend, net of taxes). The compounding inside the investee was real and substantial. The accounting was failing to record it.

Why retained earnings can be more valuable than dividends
Buffett's 1989 framing addressed a question that confuses many investors: if a company holds a stake in another business and does not receive the cash, why does that retained portion have real value? The answer: "You must judge whether these undistributed earnings are as valuable to you as those already reported. We believe they are — and, in fact, may even be more valuable. The reason is that our investees are employing their retained earnings under the direction of managers who are talented and shareholder-oriented, and who sometimes find uses for capital within their own businesses that are superior to anything we could find for it."
“
Key takeaway
Buffett's 1989 framing of the connection: "For Berkshire's intrinsic value to grow at an average of 15% per year, our look-through earnings must also grow at approximately the same rate." Look-Through Earnings is the cleanest single-number proxy for the underlying compounding engine when standard GAAP earnings do not capture the full picture. The metric matters whenever you analyze any company holding substantial non-controlling equity stakes.
The 1982 critique of the 20% threshold
Buffett challenged GAAP's arbitrary 20% threshold as early as the 1982 letter: "The value of retained earnings to all shareholders depends on the efficiency with which they are deployed, not on the size of your ownership stake." The accounting rule treats an 8% holder and a 25% holder as fundamentally different even though both have proportionate economic interest in the underlying business. The threshold creates discontinuity where economic reality is continuous, and Look-Through Earnings restores the continuity.
Practical application

Applying these metrics on Dividend Line

The metrics in this chapter are exactly the kind of second-order analysis that benefits most from an integrated data platform. Six features connect directly:

Owner Earnings approximation in Summary tab

The summary tab shows net income, D&A, total capex, and an estimated maintenance/growth split based on industry-specific ratios. Owner Earnings is calculated from these inputs and displayed alongside reported net income. The two numbers shown together reveal how much of reported earnings is economically distributable.

ROIC over 10 years

The metrics history endpoint shows ROIC year by year for the past decade. Look at the trend line, not the latest number. Sustained above 15% with stable or rising trajectory is a moat in operation. Sustained above 20% is exceptional. Declining ROIC over multiple years is moat erosion.

Cash conversion 5-year cumulative

The platform calculates rolling five-year cumulative net income versus five-year cumulative operating cash flow. The two should converge. Material divergence is the cleanest available signal of earnings quality issues.

Balance sheet stress test

The financials tab shows debt/EBITDA, interest coverage, and maturity profile. The stress test simulator (premium tab) models a 30% revenue decline against debt service requirements and surfaces businesses where the safety margin is too thin.

Earnings quality red flags screener

The screener includes filters for the eight red flags from this chapter — receivables outpacing revenue, OCF below NI, heavy adjusted earnings, recent auditor changes, insider selling. Build screens that exclude businesses with two or more flags simultaneously.

Look-Through analysis for holding companies

For holding companies and businesses with substantial minority equity stakes (Berkshire, Markel, certain insurance and asset-management companies), the platform computes Look-Through Earnings using disclosed proportionate ownership and investee earnings. Particularly relevant for analyzing Berkshire itself.
Reader questions

Frequently asked questions

Start at net income on the income statement. Add back depreciation, depletion, and amortization (D&A) — these are non-cash charges that GAAP recorded against earnings but did not actually leave the company as cash. Then subtract two things: the cash needed each year to maintain competitive position (maintenance capex) and any required increase in working capital (increases in receivables and inventory net of payables).

The hard step is splitting total capital expenditure into maintenance and growth. Three approaches: management sometimes discloses this directly in the 10-K or earnings calls (rare but valuable when present); for asset-light businesses, maintenance capex tends to track depreciation closely; for asset-heavy businesses, you may need to estimate from industry rules of thumb (typically 1.0×–1.5× depreciation for steady-state operations). Imprecise but workable. The point is to be roughly right, not exactly right.

Next lesson ◆

Chapter 6 · Valuation & Capital Allocation

Three methods to estimate intrinsic value, margin of safety tiers, the five capital allocation paths, when buybacks create or destroy value, dividends versus retained earnings, arbitrage, bonds, and convertibles.
Continue
Educational disclaimer · This content is for educational and informational purposes only. It does not constitute investment advice, tax advice, legal advice, or a recommendation to buy or sell any security. Always conduct your own research before making investment decisions.