Not All ROE Is Created Equal

Return on Equity is one of the most quoted metrics in finance. Analysts screen for it. Boards report it. Investors use it to rank companies.

But here is the problem: most people treat ROE as a single number. In reality, there are three distinct versions, and each one tells a completely different story.

Understanding which version you are looking at is the difference between spotting a great capital allocator and falling for a temporary earnings spike.

1. Average ROE: The Rear-View Mirror

Average ROE is the classic formula:

Average ROE = Net Income / Book Equity

This is the number you see in every annual report and stock screener. It represents the blended return from all capital ever deployed by the company, both old projects and new ones.

The problem? Average ROE is slow to change. It is dominated by legacy assets and historical decisions. A company could be making terrible new investments today, but if the old business is still generating solid returns, the average masks the deterioration.

Think of it like a batting average across an entire career. It tells you where someone has been. It does not tell you how they are performing right now.

2. Marginal ROE: The Front-View Mirror

Marginal ROE (also called Incremental ROE) is the metric that separates good analysts from great ones:

Marginal ROE = Change in Net Income / Change in Book Equity

This measures the return generated by new capital deployed in the most recent period. It is a forward-looking signal that reveals whether management is allocating capital wisely or destroying value.

The critical test is simple: if Marginal ROE is below the cost of equity, every new dollar the company retains destroys shareholder value. Full stop.

David Einhorn highlighted this concept powerfully in his 2006 speech. He showed that major investment banks were reporting headline ROE of 15 to 20%, which looked healthy on the surface. But the incremental returns on new capital were terrible. Approximately 50% of revenues were going to employee compensation, and the banks were reinvesting in high-risk proprietary trading rather than returning cash to shareholders.

Einhorn concluded that these banks were “run for their employees, not their shareholders.” Within two years, the 2008 financial crisis proved him right.

The lesson: headline ROE can lie. Marginal ROE tells the truth about where a company is headed.

3. The Changing ROE Effect: The Temporary Speed Boost

This is the most misunderstood concept of the three. When ROE rises from one period to the next, EPS growth temporarily spikes even without any change in the reinvestment rate. Conversely, when ROE falls, EPS growth slows more than expected.

The math behind this is the sustainable growth formula:

g = ROE x b (retention ratio)

When ROE is stable, this formula holds cleanly. But when ROE is changing, the actual EPS growth rate deviates from the sustainable rate. A rising ROE adds a temporary boost on top of the sustainable growth. A falling ROE subtracts from it.

Once ROE stabilizes at its new level, growth reverts to the sustainable rate. The spike was never permanent.

This is where many junior analysts and even experienced investors get misled. They see a year of 25% EPS growth driven by improving ROE and extrapolate it forward. In reality, the growth rate will normalize once the ROE improvement plateaus. Investment bankers know this. They normalize it in their models. But the narrative often sticks longer than the math supports.

Real-World Evidence: TVS Motor

The TVS Motor case study from FY18 to FY25 illustrates this framework perfectly. From FY18 to FY20, both ROCE and ROE declined steadily, with ROCE falling from 18.2% to negative territory. Incremental returns lagged headline metrics, and the stock price halved.

The turnaround began in FY21. As ROCE and ROE rebounded, incremental ROCE and ROE consistently exceeded headline metrics, signaling that new capital was being deployed effectively. The market responded with a dramatic rerating: the stock soared more than six times from its 2020 lows by FY25.

The investors who tracked only Average ROE missed both the decline and the recovery. Those who watched Marginal ROE saw the signals early.

The Practical Framework

Here is how to apply the three versions of ROE in your analysis:

The Bottom Line

A spike in ROE looks like a permanent story. In reality, bankers normalize it. The best investors do too.

Next time you see a company boasting about rising ROE, ask two questions: Is the Marginal ROE above the cost of equity? And is the ROE improvement sustainable, or just a one-year efficiency bump?

The answers will separate the compounders from the value traps.

Sources: David Einhorn 2006 Value Investing Congress speech; Berkshire Hathaway 1992 Shareholder Letter; TVS Motor FY18-FY25 financial data via Pro-Setups Newsletter (Jun 2025). Analysis by decoded.finance.