The single most common beginner question in value investing: what PE ratio is cheap? The most common beginner answer is some version of "below 15" or "below 10." That answer is wrong, and the reason it is wrong is the most important lesson in valuation.
There is no universal PE ratio that defines "cheap." A PE of 15 in one sector is a screaming bargain. A PE of 15 in another sector is fairly priced. A PE of 15 in a third sector is dangerously expensive. This post walks through three real sectors with worked examples to make the point unmistakable.
Why a single PE threshold cannot work
The PE ratio is price / earnings per share. The inverse, earnings yield (1 / PE), is the percentage return you would get if all the earnings were paid out as a dividend immediately. A PE of 10 means earnings yield of 10 percent; a PE of 20 means 5 percent; a PE of 33 means 3 percent.
The reason a single PE threshold fails: the question is not "what is the earnings yield" but rather "what does the earnings yield need to be, given the structural characteristics of this business?" Three characteristics dominate:
Characteristic 1: Growth rate. A business growing earnings 20 percent per year can justify a much higher PE than a business growing at 2 percent per year, because the future earnings will be much larger in absolute terms.
Characteristic 2: Capital intensity. A business that requires heavy capex to maintain its earnings (utilities, miners, integrated oil) generates lower free cash flow per dollar of earnings than a capital-light business (software, asset managers). The same PE means much higher implicit FCF yield for the capital-light business.
Characteristic 3: Predictability. A business with stable, predictable earnings (consumer staples, regulated utilities) justifies a higher multiple than a cyclical business with volatile earnings (commodities, autos, semiconductors), because the predictability lowers the required risk premium.
Three sectors, three different "cheap PE" answers.
Sector 1: Software (capital-light, high growth, decent predictability)
Example: Adobe (ADBE). Adobe runs creative software (Photoshop, Premiere, After Effects) and document software (Acrobat) on subscription pricing. Capital intensity is low (software development costs but minimal physical capex). Growth is strong (10-15 percent revenue CAGR over the last decade). Predictability is decent because the subscription model produces recurring revenue with high retention.
Adobe's PE in May 2026: roughly 36. By the "PE below 15 is cheap" beginner rule, Adobe is expensive.
But:
- Adobe's FCF margin is 35 percent (every dollar of revenue produces 35 cents of free cash flow). The same PE implies a much higher FCF yield than for capital-intensive businesses.
- Adobe is growing earnings at roughly 14 percent per year. The PEG (PE divided by growth rate) is 36 / 14 = 2.6, which is above Lynch's 1.5 threshold, but the business quality is exceptional.
- The recurring-revenue subscription model means a year of bad sales does not destroy the next year's revenue. Predictability is high.
For software at Adobe's quality and growth, the historical fair-value PE range has been 25 to 45. A PE of 36 is in the middle of that range. Adobe is fairly priced for what it is, not expensive.
A PE of 15 in software (say, Oracle in some periods) would be an exceptional bargain by the same logic, because software earnings are higher quality per dollar than utility earnings.
Sector 2: Regulated electric utility (capital-intensive, low growth, very high predictability)
Example: Duke Energy (DUK). Duke is a regulated electric utility serving the Carolinas, Florida, and a few other Southeast markets. Capital intensity is enormous (multi-billion-dollar power plants, transmission lines). Growth is low (roughly 4-6 percent earnings growth as a long-term ceiling, mostly through approved rate-base increases). Predictability is exceptional because the regulator sets allowed returns.
Duke's PE in May 2026: roughly 19. By the beginner rule, Duke is at the edge of fair value.
But:
- Duke's FCF after capex is structurally lower than EPS because of ongoing capital reinvestment in grid modernisation and renewables. The reported PE of 19 corresponds to a free-cash-flow yield in the 3-4 percent range, not 5 percent.
- Earnings growth ceiling is around 5 percent. PEG = 19 / 5 = 3.8, which would look terrible by Lynch's rule, except utilities are not Lynch candidates by design.
- The 5 percent dividend yield is the actual cash return investors get.
For a regulated utility, the historical fair-value PE range is 15 to 20. Duke at 19 is fairly priced to slightly stretched, but not catastrophically expensive. A PE of 15 would be a bargain. A PE of 25 would be expensive enough that the dividend math no longer makes sense.
Notice the boundary: the same PE of 19 that means "fairly priced" for Duke means "deep bargain" for Adobe. The sector context completely changes the interpretation.
Sector 3: Coal mining (capital-intensive, no real growth, low predictability)
Example: Peabody Energy (BTU) at a cyclical peak. Peabody mines and sells thermal and metallurgical coal. Capital intensity is high (mining equipment, environmental remediation reserves). Growth is essentially zero in real terms; the industry is in long-term secular decline as power generation shifts away from coal. Predictability is poor because earnings swing with coal prices.
Peabody's PE in May 2026, near a cyclical peak: roughly 6. By the beginner rule, Peabody is a screaming bargain.
But:
- The "E" in PE is at a cyclical peak. Coal earnings can swing from +5 billion to -1 billion across a single cycle. Peabody at PE 6 on peak earnings has a normalised PE that is much higher.
- The earnings are not growing; they are reverting to mean. Buying at peak earnings on a cyclical means buying the top of the cycle.
- Long-term, the addressable market is shrinking. Coal demand globally peaks somewhere between 2027 and 2030 by most forecasts; the long-run trend is negative.
For a peak-cyclical commodity at a cyclical top, the historical pattern is that PE looks deeply cheap immediately before earnings collapse. A PE of 6 on peak earnings is not cheap; it is expensive on the normalised earnings power of the business. A PE of 20 on trough earnings might actually be cheaper.
The PE ratio is a near-useless signal for peak-cycle commodities. Better metrics: price to normalised mid-cycle earnings, or price to long-run replacement cost of the productive assets.
The general principle: match the PE to the business
The honest rule for using PE in value investing is to match the threshold to the structural characteristics of the business:
| Sector type | Reasonable PE range | Why |
|---|
| Quality software / platforms | 25 to 40 | High FCF conversion, durable growth, predictable revenue |
| Wide-moat consumer staples | 18 to 28 | Stable demand, strong brand, lower growth than tech |
| Regulated utilities | 14 to 20 | Predictable but capital-intensive, low growth |
| Banks and insurers | 8 to 14 | Earnings depend on interest-rate environment and credit cycle |
| Mature industrials | 14 to 22 | Cyclical but with brand or franchise quality |
| Commodity producers | use price to normalised mid-cycle earnings, not PE | Cyclical earnings make PE deceptive |
| Pharma | 15 to 22 | Patent cliffs and pipeline risk dampen the multiple |
| REITs | use FFO or AFFO multiples, not PE | Real estate depreciation creates misleading GAAP earnings |
These ranges are not laws of physics; they are observed historical centres for typical businesses in each category. A specific business can deserve a higher multiple if its quality is exceptional or a lower multiple if its quality is impaired.
How to use PE correctly
A practical 3-step approach:
Step 1: identify the sector category of the business.
Step 2: compare the current PE to the historical fair-value range for that sector category.
Step 3: check whether the business is at the top or bottom of its sector range. If it is at the bottom of the range AND quality is high, it is likely cheap. If it is at the top of the range AND quality is impaired, it is expensive.
Step 4 (the most important): ignore the PE entirely for cyclicals. Use normalised earnings or replacement-cost metrics.
Why owner earnings yield is usually a better signal
For most businesses, the owner earnings yield (computed as owner earnings divided by market cap) is a cleaner cross-sector metric than PE, because owner earnings adjust for the capital-intensity differences between sectors. We walk through owner earnings in detail at /blog/what-is-owner-earnings-buffett/ and the broader valuation triangulation at /blog/how-to-value-a-stock-2026/.
A 5-7 percent owner earnings yield is the rough "fair price for a wonderful business" threshold across most sectors. Below 3 percent is stretched. Above 8 percent is cheap. The threshold is much more sector-stable than the PE threshold.
Where invest-like fits
Every stock page on invest-like surfaces both the PE ratio and the owner earnings yield, with the sector context overlaid. The Buffett-Fit Score's Valuation pillar uses owner earnings yield as the primary signal precisely because it is more sector-comparable than raw PE.
For sector-specific rankings:
The takeaway in three sentences
-
There is no universal "cheap" PE. The right threshold depends on the sector's growth rate, capital intensity, and earnings predictability.
-
A PE of 15 is a bargain in software, fairly priced in regulated utilities, and a warning sign in cyclical commodities at their peak.
-
Owner earnings yield is a cleaner cross-sector metric for "am I overpaying" than raw PE.
Disclosure
Educational tool. The PE ranges in the table above are observed historical centres, not regulatory thresholds. Specific stocks can deserve different multiples. The named companies (Adobe, Duke, Peabody) are illustrative and not investment recommendations. Past PE ranges do not predict future PE ranges.
Author: Zaid Ghazal, founder of invest-like, Kiel, Germany. Not a registered investment adviser. Full methodology at /methodology/.