Some pie charts, of questionable utility
The Economist has another example of a graph that tells the viewer, well, nothing.
Last time it was due to comparisons between data sets which weren’t really comparable. This time, it’s an odd choice of data to start with.
Apparently, trade diversity is important. My study of economics did not go beyond A-level, but I’m inclined to agree; common sense certainly implies that a country producing a narrow range of products is not going to have a very stable economy. Measuring trade diversity is probably slightly tricky. Off the top of my head, I would suggest looking at the number of exporting industries in a country, perhaps compared to the number of importing industries, and maybe scaled for land or population size. I’m not an economist, but something like that would make sense to me.
This chart, however, assumes that trade diversity can be measured by looking at how much of a country’s total exports are composed of the top five exported products. It implies that the lower this figure, the more diverse the trade.
By this measure, Italy has the most diverse trade. In fact, of the top ten countries (those for which there are pie charts), seven are in the EU. The other three are Croatia (a candidate for EU membership), the USA and China.
The first noticeable issue with the graph, however, is that it gives no indication of these countries’ export market sizes – which I would have thought would be important when making comparisons between, for example, China and Romania. It proves difficult to find actual figures for those two countries on the internet, but I would guess that they are slightly different.
The next obvious problem with the data is how the categories of imports and exports are broken down. The top five imports for each country are listed. But “cars” and “car parts” are in different categories, it seems – and neither includes “engines”, which are different again. “Outer garments”, “jackets” and “trousers” are similarly distinct.
Clearly, dividing goods up into different categories will sometimes be a little arbitrary. But it makes sense to assume that these categories will be the same for each country, to facilitate comparisons. But then why does the list for Poland differentiate between “chairs” and “furniture parts”, while Croatia has an entry for “chair parts”? Either there is inconsistency in labelling (which is bad enough) or the charts are not comparing like with like, rendering analysis all but meaningless.
More troubling is the lack of units. Are the figures for value? Profit? Weight? Containers (or ships) filled? At a guess, I would say value is the most likely answer – but if it is not, then that would skew the data. The lack of units immediately leads me to question the reliability of the graph. Romania exports a lot of electric cables, but if the criteria is tonnage then it is an even more impressive figure when compared with the US’s aeroplanes. Units are important here. But try as I might, I couldn’t find relevant data online to figure out what figures are being used. Perhaps I’m just not looking hard enough.
And then, as if that wasn’t enough, there is the problem with using these statistics in comparison.
Consider hypothetical countries X and Y. The Republic of X produces exports in six categories, with the trading figures (we will assume in terms of value) split equally – each category accounts for approximately one-sixth of the total export market. So the percentage of the market accounted for by the top five exports is 83.3%. But the Kingdom of Y produces exports in eleven categories. The top five categories make up one sixth of the total market each, with the remaining sixth split equally between the last six categories. So the same figure – 83.3% – applies to both countries, but clearly they do not have the same level of trade diversity. And this is assuming that both countries have the same total value of exports.
Consider two more countries, W and Z, both producing 15 different categories of exports. W has exports totalling 20AMUs (Arbitrary Monetary Units), while Z’s total 40AMUs. Suppose both countries produce 1AMU worth each of export categories (1) through (10). But W produces 2AMUs’ worth each of the remaining categories (11) through (15), while Z produces 6AMUs’ worth of each.
In this situation (which would be clearer if I could work out how to insert charts – I apologise), the relevant figures (for the top 5 exports) are 50% for W and 75% for Z. Are we to conclude that Z is thus less diverse?
As I’ve said before, I’m not an economics scholar. But it strikes me that this graph does not even come close to giving the full picture in measuring trade diversity. And that’s before considering whether this statistic is even a good measurement for diversity. Shouldn’t the number of countries that a given country trades with be considered, too? Or the number of industries exported to, as well as from? Not to mention whether exports are raw materials, components or finished products. In different contexts, all these could be important.
I realise that The Economist‘s Daily Chart feature is just trying to present a snapshot of relevant figures in an accessible form. And it usually manages it quite well. But sometimes, it just seems to be misleading. Maybe the writer in charge should step away from the pretty graphics software, and take a moment to flick through this.
In fact, anyone with a vague interest in graphs should do so too. It’s well worth a glance.
Filed under: Comment on Comment, Mathematics & Statistics | Leave a Comment
Tags: Exports, Graphs, Mathematics & Statistics, Misleading Statistics, The Economist, Trade