Fisher’s Index is a price, quantity, or value tracker economists love because it splits the difference between two classic methods. It’s basically the geometric mean of the Laspeyres and Paasche indices, which gives it a nice balance—and the U.S. Bureau of Labor Statistics still swears by it in 2026 for its accuracy and reversibility.
Quick Fix Summary
To calculate Fisher’s Index:
- Run the Laspeyres Index with base-period quantities.
- Run the Paasche Index with current-period quantities.
- Take the geometric mean of both results to land on Fisher’s Index.
What Is Fisher’s Index?
Fisher’s Index is the geometric mean of the Laspeyres and Paasche indices, designed to cut the bias that comes from using only base- or current-period quantities.
Economist Irving Fisher cooked this up in 1922 to fix problems with older fixed-weight indices. Instead of locking in base-period quantities like the Laspeyres Index or chasing current-period quantities like the Paasche Index, Fisher’s Index blends both, evening things out. The International Monetary Fund points out it passes two key tests—time reversal and factor reversal—so it’s great for long-term comparisons.
Step-by-Step Solution: Calculating Fisher’s Index
To compute Fisher’s Index, first calculate the Laspeyres and Paasche indices, then take their geometric mean.
Here’s how to do it for prices across two periods—Year 0 (base) and Year 1 (current).
- Compute the Laspeyres Price Index
Plug in base-period quantities (q0) and both base- and current-period prices (p0 and p1):
Laspeyres Index = (Σ p1q0) / (Σ p0q0) × 100
(Example: Σ p1q0 = $120, Σ p0q0 = $100 → Laspeyres Index = 120.)
- Compute the Paasche Price Index
Now use current-period quantities (q1):
Paasche Index = (Σ p1q1) / (Σ p0q1) × 100
(Example: Σ p1q1 = $130, Σ p0q1 = $110 → Paasche Index = 118.18.)
- Calculate the Geometric Mean
Fisher’s Index = √(Laspeyres Index × Paasche Index)
(Using the numbers above: √(120 × 118.18) ≈ 119.08.)
If This Didn’t Work
If your Fisher Index result feels off, try a Chain Fisher Index, Marshall-Edgeworth Index, or switch to a fixed-basket method.
Something feel wrong? Here are three ways to fix it:
- Use a Chain Fisher Index — For multi-period work, link individual Fisher indices by multiplying consecutive values and dividing by 100. That’s exactly what the U.S. Census Bureau does in its data releases.
- Try the Marshall-Edgeworth Index — This one splits the difference by averaging base and current quantities as weights. Formula: (Σ p1(q0+q1)) / (Σ p0(q0+q1)) × 100.
- Switch to a Fixed-Basket Method — When data jumps around, stick with a fixed basket (Laspeyres) but refresh weights every year to keep bias in check, as the OECD suggests.
Prevention Tips: Avoid Index Bias
To keep Fisher’s Index honest, update quantities regularly, cross-check with other methods, and document every data source.
Want clean results? Follow these three habits:
- Update Quantities Regularly — Rotate base periods every five years or so to stop old weights from warping your numbers.
- Validate with Multiple Methods — Run Fisher’s Index alongside Tornqvist or Walsh indices—they also use geometric averaging and are the go-to for modern CPI work.
- Document Data Sources — Note where your prices and quantities come from (BLS, Eurostat, etc.) and flag any substitutions or imputations you made.
| Index Type | Formula | Weight Basis | Best Use Case |
|---|---|---|---|
| Laspeyres | (Σ p1q0) / (Σ p0q0) × 100 | Base-period quantities | Long-term trend analysis with stable consumption |
| Paasche | (Σ p1q1) / (Σ p0q1) × 100 | Current-period quantities | Short-term inflation tracking with changing consumption |
| Fisher | √(Laspeyres × Paasche) | Average of base and current quantities | Comparative studies requiring reversibility and symmetry |
Since Irving Fisher published his index in 1922, it’s earned its reputation as the “ideal” measure thanks to its rock-solid math. Sure, newer indices like the Chained CPI now show up in some U.S. federal programs, but Fisher’s method is still the backbone of economic research and teaching.
