A sector P/E depends on which companies and earnings you combine

An average company P/E and a sector's combined market value divided by combined earnings are different statistics. Both may appear under a short sector-multiple label, but they weight companies differently. Before comparing a business with either number, establish the calculation and the population included.

Damodaran's NYU sector table distinguishes multiple earnings bases and includes separate aggregate columns for all firms and money-making firms. The labels expose choices that a single sector P/E can otherwise hide. This guide uses fictional companies to show the arithmetic without relying on a current market ranking.

Record whether the intended question concerns a typical company, the combined economics of a group, or a particular weighted portfolio. That purpose helps determine the statistic. A median, an arithmetic mean and a ratio of totals should not be interchanged merely because they are all expressed as multiples.

Suppose fictional Company A has an equity market value of $1,000 million and annual earnings of $100 million, producing a P/E of 10 times. Company B has a market value of $100 million and earnings of $1 million, producing a P/E of 100 times. Assume the earnings and values use consistent dates and definitions.

The equally weighted arithmetic average of the two company multiples is 55 times. The aggregate calculation instead divides $1,100 million of combined market value by $101 million of combined earnings, giving approximately 10.89 times. The large difference is not a rounding error; the two calculations answer different questions.

In the arithmetic mean, Company B receives the same influence as Company A despite being much smaller by market value and earnings. In the ratio of totals, the underlying amounts determine the result. Keep those amounts visible so the reader can see why a high multiple attached to a small earnings base affects the statistics differently.

A market-value-weighted arithmetic average of the company P/E ratios is another calculation. For the same two companies, the weights are 1,000/1,100 and 100/1,100. Applying them directly to 10 and 100 produces approximately 18.18 times, which still differs from the aggregate 10.89 times.

For positive earnings, the aggregate ratio can instead be reproduced by combining earnings yields using those market-value weights and then taking the reciprocal. The yields in the example are 10% and 1%. Their weighted yield is approximately 9.1818%, whose reciprocal is approximately 10.89 times.

This identity is a useful calculation check, not a reason to obscure the simpler ratio of totals. Retain the combined market value and earnings as the primary reconciliation. If a data provider describes a weighted multiple, inspect whether it uses an arithmetic average, a harmonic construction or another methodology.

Use full-precision inputs for the check. Averaging rounded displayed multiples can introduce another difference, particularly where one company's earnings are small. Preserve the source amounts and rounding policy rather than adjusting the total to match a rounded summary table.

Add fictional Company C with an equity market value of $200 million and a $20 million annual loss. Including all three companies gives aggregate market value of $1,300 million and aggregate earnings of $81 million. The ratio of totals is approximately 16.05 times.

Excluding Company C from both the numerator and denominator returns the profitable-company aggregate to approximately 10.89 times. Neither number should be labeled as the other. The all-company measure includes the loss and the related market value; the profitable-company measure describes a narrower population.

A mismatched exclusion is a separate error. Removing Company C's loss from earnings while retaining its $200 million market value produces $1,300 million divided by $101 million, or approximately 12.87 times. That figure does not match either consistently defined population.

Keep a company-level inclusion flag and reconcile both totals to the same selected rows. Record the reason for exclusions such as missing data, negative earnings or an out-of-scope business. The population is part of the measure and should survive export into a presentation.

Combine earnings on a common stated basis, such as a defined trailing period or a specified forecast horizon. Mixing historical earnings for one company with forward earnings for another changes the aggregate in a way that is difficult to interpret. Preserve the earnings source and period at company level.

Align market-value timestamps and currencies before adding amounts. Check share classes and ownership units so that the market value corresponds to the earnings attributed to those holders. A ratio of carefully summed but incompatible amounts can still be wrong.

Review acquisitions, disposals and listing changes when comparing the sector over time. An aggregate can move because the member list changed, even if continuing members' valuations and earnings did not. A constant-population comparison can help explain that movement when it is clearly labeled and reproducible.

Retain stable company identifiers with the sector classification. Name changes or multiple share listings should not accidentally create two copies of the same earnings in the combined denominator.

Publish the statistic's name, company count, inclusion policy, earnings basis and market-value date with the multiple. For a ratio of totals, include the combined numerator and denominator. For an average or median, state the weighting and treatment of missing or unsuitable observations.

If aggregate earnings approach zero or become negative, explain why the resulting multiple is unstable or unsuitable for the usual comparison. A very large or negative number should not be presented as an ordinary sector benchmark without that context. Retain the component earnings to show what happened.

The fictional examples produce 55.00, 18.18, 10.89, 16.05 and 12.87 times from different operations on closely related inputs. A useful report identifies the intended calculation and rejects the inconsistent one. That gives the reader a benchmark they can reproduce and interpret within its stated limits.