Heterodox Economics Newsletter

Issue 361June 22, 2026 web pdf Heterodox Economics Directory

As researchers most of us supposedly like the mean. The mean is seemingly very kind to us: it is a simple number, it (supposedly) summarizes a lot of information and it can serve as (somewhat) solid anchor for further statistical reasoning. Hence, all praise to the mean and its convenience ;-)

While I think the above is largely correct, it is also incomplete: precisely because the mean compresses information into a single number, it can also hide important patterns. This matters whenever distributions are multimodal, when multiplicative dynamics are at work, or when changes in inputs produce qualitative shifts in outputs — for example, in the relation between income and subjective well-being. In such cases, a mean-based view can conceal substantial heterogeneity.

A key example of how the mean can hide significant aspects comes from a feminist perspective on measuring poverty (see here or here). Such a view emphasizes how using household means as a baseline for constructing poverty estimates conceals lack of disposable income for household members with no or marginal employment – often women. This often amounts to rendering women’s poverty invisible and, in addition, imposes a downward bias on standard poverty estimates.

Another example is given by GDP as a key economic figure, or more precisely, its growth rate. GDP growth is typically given by the growth of mean incomes, which gives the impression that the GDP growth rate resembles the „typical“ change in individual outcomes in an economy. However, this is, generally, incorrect: GDP growth measures growth in money terms. Hence, the underlying „mean growth“ does not refer directly to individuals, but weighs them by their income. As a consequence, the growth of richer individuals matters more for GDP growth than does the growth rate enjoyed by the poor.

A truly democratic mean – more apt to represent the typical experience of an individual – would require us to calculate not the growth in average income (which employs a subtle weighting and, hence, can easily be driven by a few if the distribution is unequal), but, rather, the average growth in income (see here). Only in the second calculation – where people instead of money enter the denominator, and hence distributional issues play a role* – the mean is informative about the „typical“ individual experience.

Finally, climate change adds yet another case. Mean temperature changes are very useful, but they can also create the impression of a slow and uniform process. Looking at extremes often tells a different story: across many regions, temperature extremes change faster than the mean, and climate risks are frequently driven by rare events rather than by average conditions. This is especially visible in phenomena such as heatwaves, cold spells, or El Niño events, whose impacts are often disproportionate to what mean temperatures alone would suggest.

Of course, this shift in outliers does not contradict the mean. It just points to the fact that very relevant variables in economics – like changes in urban temperature settings, but also key variables like income, wealth or firm size – intrinsically require considering extreme cases if they are to be properly assessed and understood.

In this spirit: stay cool in the upcoming heatwaves, and all the best,

Jakob

PS: If you are now inspired to think more deeply about what it could mean to do „statistics beyond the mean“ (what a pun!), I can warmly recommend to contact this fantastic and inspiring colleague of mine.

*In general, the democratic growth rate will always be greater than the standard growth rate if the distribution becomes more equal and vice versa.

© public domain

Table of contents