Allen Framework vs Buffett Owner Earnings: What Changed

I've been reading annual letters since the 90s. And if you've been around value investing long enough, you know the moment everything clicked: 1986, Berkshire Hathaway's letter, buried in an appendix titled "Purchase-Price Accounting Adjustments and the 'Cash Flow' Fallacy."

That's where Warren Buffett dropped Owner Earnings on us.

It wasn't flashy. No headlines. Just a quiet revolution: "Stop looking at what accountants tell you. Look at what actually flows into shareholders' pockets."

Simple. Profound. And — here's the thing — unfinished.

Because Buffett left us with a problem. That little (c) in the formula? "The average annual amount of capitalized expenditures for plant and equipment needed to maintain the business's competitive position and unit volume."

He literally said: you'll have to guess.

And for 40 years, that's what we've been doing.

Until now.

Why Buffett's Owner Earnings Formula Needed an Update

Let me be clear: Buffett wasn't wrong. He was vague on purpose. He quotes Keynes in that same letter (often attributed to Keynes): "I would rather be vaguely right than precisely wrong."

Fair enough. When you're Warren Buffett, you can afford to rely on intuition backed by decades of experience. You can look at a company and feel how much of its capex is maintenance versus growth.

But what if you're not Warren Buffett? What if you want to build a system that can monitor 10 companies simultaneously, update valuations in real-time as new information arrives, and — here's the kicker — actually show its work?

That's where the Allen Framework comes in.

I've spent the last 20+ years wrestling with this exact problem. Not because Buffett's philosophy was flawed, but because the world changed. We have AI now. We have structured data pipelines. We have the tools to engineer what Buffett could only philosophize about.

And we built it.

3 Key Innovations Beyond Buffett's Owner Earnings

Innovation 1: The Maintenance Ratio

Buffett said subtract (c) — the capex needed to maintain competitive position. But he never told us how much that was.

The Allen Framework says: make it explicit.

See that Maintenance Ratio? That's company-specific. For TSMC, it's 20% — because 80% of their massive capex goes toward growth (building cutting-edge fabs), not just keeping the lights on.

For a mature railroad company? Maybe 80% maintenance, 20% growth.

This isn't guesswork. It's parameterized guesswork — which means as IVCO's database grows, as we track annual reports and earnings calls and capex announcements, that ratio gets more accurate.

Why this matters: Without this, you'd look at TSMC's $1 trillion in capex and think "holy hell, their free cash flow is terrible." But that's not maintenance. That's expansion. The Allen Framework separates the two — and suddenly, you see TSMC's true earning power.

Innovation 2: The Reality Coefficient

Buffett talked about removing accounting distortions — like when Berkshire acquired Scott Fetzer and purchase-price adjustments made GAAP earnings look $11.6M worse, even though the economic substance was identical.

But he only addressed one kind of distortion: acquisition accounting.

The Allen Framework says: let's systematize all distortions.

Every year's reported Owner Earnings gets a Reality Coefficient:

• 100% = normal year, use as-is

• >100% (like 125%) = one-time loss that year, adjust upward

• <100% (like 80%) = one-time gain that year, adjust downward

When you calculate CAGR from historical data, you're not just taking the raw numbers. You're taking the truth behind the numbers.

Early-stage method: Take 3-year averages at endpoints to smooth out anomalies.

Advanced method: Assign a Reality Coefficient to every single year as your database matures.

Why this matters: A single lawsuit settlement or foreign exchange gain can throw off your growth rate by 5%. Multiply that over 10 years of compounding, and you're off by 50%. The Reality Coefficient fixes that.

Innovation 3: The Confidence Coefficient

This is the big one.

Buffett's 1986 letter was about calculating historical Owner Earnings correctly. It didn't tell you how to forecast the future.

The Allen Framework does.

We introduce the Confidence Coefficient — a multiplier that sits on top of your historical CAGR to reflect forward-looking conviction:

Let me give you a real example from early in my career.

The Case That Proved It

I found a precision component manufacturer — a Tier 2 supplier buried deep in the supply chains of some of the world's biggest brands. In their annual report — publicly disclosed, right there in black and white — they were building production capacity 3 times their current size.

Not "we're thinking about it." Not "we hope to." They were building it.

The market? Didn't care. Stock price barely moved.

I applied what's now the Confidence Coefficient. Historical CAGR was solid. New capacity was verified. Management had a 100% track record of delivering on commitments.

CC = 2.5x to 3.0x.

The intrinsic value range I calculated was so far above the market price that the margin of safety was laughable.

That's what the Allen Framework does. It takes publicly available information — the kind Fisher taught us to hunt for, the kind Buffett taught us to analyze — and quantifies it into a math problem you can actually solve.

From Owner Earnings Philosophy to a Complete Valuation Algorithm

Buffett gave us the starting point: how to calculate one year's Owner Earnings.

The Allen Framework gives us the full pipeline:

Let's run TSMC through it.

TSMC: Theory Meets Reality

Not a single point estimate. A range. Because Buffett was right: vaguely right beats precisely wrong.

But now? That vague rightness is backed by an algorithm. Every assumption is documented. Every coefficient is traceable. If new information comes in — a capex announcement, a management change, a competitive threat — IVCO updates the coefficients, reruns the model, and outputs a new range.

In real-time.

What Buffett Left Us — And What the Allen Framework Adds

Here's the thing: this isn't a replacement. It's a continuation.

Buffett wrote that letter because Wall Street was lying to itself, calling EBITDA "cash flow" while ignoring the reality that every piece of equipment eventually needs replacing.

The Allen Framework takes that insight and asks: okay, how much needs replacing? And how much of what you're spending is actually expansion?

Because in 1986, the world's greatest investors were running industrial conglomerates and insurers. Today? TSMC spends a trillion on capex to build 2nm fabs. SpaceX spends billions on reusable rockets. Tesla builds gigafactories.

If you treat all capex as "maintenance," you'll never buy a growth company.

The Allen Framework doesn't just honor Buffett's philosophy. It extends it into the 21st century.

One More Thing: The Four Masters

I didn't do this alone. Nobody does.

The Allen Framework integrates the wisdom of four giants:

• Graham: Facts and safety margins. We output ranges, not false precision.

• Buffett: Owner Earnings as the core energy. That foundation doesn't change.

• Fisher: Qualitative research. Our Confidence Coefficient comes from reading, listening, and yes — "scuttlebutt."

• Munger: Inversion and stress testing. Our CC can go below 1.0. We're not perma-bulls.

If Buffett gave us the philosophy, the Allen Framework gave us the algorithm.

And if you're a value investor in 2026, you deserve both.

What's Next

This is the first in a series where we unpack the Allen Framework piece by piece. Next up:

• TSMC Case Study: How we turned 10 years of financials into a NT$4,565–$5,639 intrinsic value range

• The Confidence Coefficient Deep Dive: What drives it, what caps it, and why one hidden champion got a 3x multiplier

• Why We Built IVCO: The story behind the system

If this resonates with you — if you've been waiting for someone to turn Buffett's footnotes into running code — welcome to IVCO.

Intrinsic value you can trust. Backed by facts. Updated in real-time.