Rural Livelihood Diversification Index

They are two sides of the same coin. The Herfindahl index is the sum of squared income shares: it is a concentration measure, so it is high (near one) when one source dominates and low when income is spread across many sources. The Simpson diversity index is its complement, one minus the sum of squared shares, so it runs the other way — higher means more diversified. Most livelihood studies report the Simpson version because 'higher equals more diverse' is more intuitive, but the two carry identical information. The inverse of the Herfindahl also gives the effective number of income sources.

No, and this is the key interpretive caution from Ellis. Diversification has two faces: accumulation, where a household adds activities to seize profitable new opportunities, and coping or survival, where it scrapes together marginal activities because farming alone cannot sustain it. Both produce a high index. So a diversified portfolio can signal a thriving, opportunity-rich household or a vulnerable one patching together a living. The index measures the extent of diversification, not its cause, so it must be read alongside the household's asset base, the quality of its activities, and the motivation behind them.

Categories should be mutually exclusive, collectively exhaustive, and matched to the research question, and they must be applied identically across all households so indices are comparable. A common scheme separates crop income, livestock income, agricultural wage labour, non-agricultural wage labour, self-employment or business, remittances, and transfers. Too few categories will mask real diversification (e.g., lumping all off-farm work together), while too many will inflate it artificially. Because the index's possible maximum depends on the number of categories, you should either normalise the Simpson index by its maximum or report the effective number of sources when comparing across studies that used different schemes.