Forest Index
Fix what qualifies first, then talk about an index
An index only means something if there are enough verifiable real samples. Today the sample count is zero — registration is not even open. Publishing any "index value" under those conditions is fabrication. So this page sets the rules instead: what data qualifies to enter an index, and what is still missing.
0
Three bottom lines
- Samples below the verification threshold do not enter the calculation. "Operator-declared" and "third-party verified" cannot be averaged together — mix them and the index measures willingness to fill in forms, not forest value.
- Samples with inconsistent bases do not enter the calculation. Stumpage and delivered prices can differ twofold; mean annual increment and current annual increment can differ widely. One index can carry only one basis.
- Groups without enough samples are not published. Better to leave a gap than to publish a "for reference only" number — once such a number is cited, it cannot be recalled.
Admissibility thresholds
The thresholds below are executable conditions, not adjectives. If any one fails, that group is not published.
| Threshold | Requirement | Why set this way |
|---|---|---|
| Verification level | Only "surveyed" and "third-party verified" samples enter the index; "claimed" samples are counted separately | Averaging data of different reliability lets the weakest part decide the conclusion |
| Minimum sample size | At least 30 independent samples per published group, of which at least 10 third-party verified | With too few samples, one or two unusual transactions move the whole number |
| Recency | Price and cost samples within the last 24 months; growth samples may extend to 60 months | Timber prices move; mixing in older data misstates the current level |
| Unit basis | One unit per group (mu/hectare, RMB/m³, USD/m³); converted samples must retain their original values | Wrong conversion factors are the most common contamination and are hard to detect afterwards |
| Price basis | Stumpage and delivered prices published separately, never merged | The gap between them is itself important information |
| Growth basis | MAI or PAI must be stated; unlabelled samples are not used | An unlabelled figure cannot be interpreted |
Grouping dimensions
An index cannot publish one "global average" — that is useful to no one. Groups must be published alongside, or nothing is comparable.
- Country / regionAt least by country; major producers split into sub-regions
- Species / cloneSpecies must be separate; note clone generations where they differ greatly, as with eucalyptus
- ObjectiveSmall/medium/large sawlogs, pulp, panels — rotation and price all change with it
- Age classSegmented by age class; never averaged across ages
- Tenure and intensityExtensive collective management separate from intensive state management (measured gaps above 60% in one production area)
- Verification levelMarked as its own dimension, never merged
Outlier treatment
Skip this and a few extreme samples drag the index. The rule must be fixed in advance — data cannot be selected after the fact.
- Trim by quantile: keep the 5%–95% range, remove the rest and record how many and why, publishing that alongside the index.
- The removal log must be public: who was removed, and why. Otherwise "outlier treatment" quietly becomes "pick the flattering data".
- Do not use the mean as the headline: publish the median and interquartile range; the mean is reference only.
- Below the sample threshold, publish no statistics at all — including no median.
Current sample status
Stated plainly — every figure here is zero, and that zero is the most important information on the page.
| Item | Today | Needed to publish |
|---|---|---|
| Forest IDs registered | 0 | ≥ 30 per group |
| Of which third-party verified | 0 | ≥ 10 per group |
| Usable real transactions | 0 | ≥ 30, traceable |
| Countries / regions covered | 0 | at least 3 major producers |
| Index values published | 0 (we publish none) | only after samples qualify |
Metadata every release must carry
Whether an index can be cited depends on whether it explains how it was computed. All of the following are required.
- Sample size, verification mix, regions and time window covered;
- Units and price basis used (stumpage or delivered; MAI or PAI);
- How many samples were removed and why;
- Median and interquartile range, plus an explanation of any change in sample size;
- The method and the date of any rule change — rules may only be changed in advance, never after the fact.
Basis
- We publish no index values at all today. This page is the admissibility rule, not an index; with zero samples, publishing any number would be fabrication.
- Rules exist before the data does: any change after publication must carry a date and a reason, so rules cannot be bent to produce a desired result.
- An index is a statistical description, not a valuation opinion, and cannot be used directly as a basis for collateral, transactions or expropriation compensation.
- Sampling and use must comply with personal-information and cross-border data rules; confidential or privacy-sensitive forest information does not enter index samples.