Wednesday, January 25, 2012

A veil of tiers

I am wrestling with statistics once again.

I'm trying to write a piece on domestic consumption profiles in China, and the data my employers have given me to work with is all over the place. For example, one stat - a survey on the new importance of 'healthy lifestyles' to better-off Chinese consumers - suggests that 4.5 million people can now be considered as displaying this preference; this, I am told, represents "1 in 16 people in China". Er, not if you accept the generally agreed population total of 1.3-and-a-bit billion, it doesn't. It's barely 1 in 300. What gives?  (There is, alas, very little chance that even my laowai analyst colleagues would pause to query something like this, and absolutely none at all that any of the Chinese researchers would.)

My main concern, however, is not with identifying and quantifying specific consumer segments (we all know those kinds of numbers are just made up, anyway; although it is nice if our inventions can be at least vaguely plausible), but with the broad question of the urban/rural population split - what proportion of of the Chinese are still living "in the countryside", and how is the urban/rural distinction defined?

Almost everything you read about China's urban centres refers to a 'tier' classification. Unfortunately, this nomenclature is very slippery, and there is no standard definition of what the tiers are. The most thorough discussion I could find of the issue is here; but even these suggested criteria still leave considerable uncertainties - with occasionally vast divergences as to how many cities and which ones fall within the 1st and 2nd Tiers - because they don't take into account the problem of how one gauges the population (or the geographical size, or the GDP, or any other worthwhile metric) of a given city. City governments invariably control great swathes of territory far beyond their core urban area, and most population figures (etc.) quoted will usually be for that entire area, including outlying suburbs, satellite towns, expanses of open countryside; the population of the city proper is probably rarely more than 50% of the headline figure, and perhaps often less than 25% - but it depends where you draw the line.

So, allocating cities within the upper tiers and assessing their size/wealth is a thorny problem. But I am more concerned with the lower end - the so-called Tier 5 cities, communities still so comparatively bijou that they might not be considered as "urban" at all (although they are sure as hell a lot more "urban" than an isolated farmstead or a mountain village of a few dozen people - which are the kind of places where a lot of Chinese people [15%??] still live).

The data supplied by the management consultancy I'm doing this article for alleges that this Tier encompasses more than 20,000 settlements, with a total population of around 150 million people.  Hmm - that doesn't sound like cities to me.

Indeed not. No, there is a recognised cohort of 'small cities' of much larger population than this, and which most lists number at around 500 (yes, that's in addition to the 42 cities that, even by the most conservative assessment of 'core urban population', number well over 1 million people; less finicky catalogues suggest that there are now more than 100 Chinese cities of this size; and then there are another 50 or so bubbling under with populations in the high hundreds of thousands). And these very numerous 'small cities' are commonly said to accommodate not a mere 11.5% of the country's population, but a far greater proportion, something around 25-30%.

In this article, for instance, I found a table of information on the city Tiers, said to be derived from the '2008 China City Statistical Yearbook' (no publisher cited). This repeats the commonly quoted figure of 494 cities in the "small, undeveloped" 5th Tier category (since we're now three years on, you can probably bump all of these figures up by at least 10%). However, certain aspects of these figures are very funny - funny peculiar, that is. In particular, the "median population" of these Tier 5 cities is said to be 609,000. Now, it would be very odd to give the median figure for this kind of thing; and you'd expect it to be lower than the arithmetic mean (because only 10 or 20% of cities in this category have attained a higher-than-average population size). I suspect this is a clumsy translation from a Chinese source, and that 'mean' was intended; although I suppose it is just as possible that it was the original Chinese survey that made this mistake. 

I was initially sceptical about that 600,000 figure - but I was myself suffering the Chinese curse of misplacing a decimal place. I suppose that would give a total of 300 million for people living in this tier - which sounds about right (not 3 billion, as it had seemed to me at first!!).

However, it still doesn't solve my fundamental problem: are these Tier 5 'small cities' considered urban or rural?


John said...

I don't understand what this task entails. They've given you a bunch of rough data, the accuracy of which doesn't seem important and you're to write a report on this? You're not a statistician so they can't want graphs and tables and the like so I can only assume they want a rough report (in English), the accuracy of which doesn't seem important? I'm not saying that would suffice but I'm at a loss as to what they want you to actually do. Do you often get back to your employers either asking for better data or with their sums actually done properly (with the decimal places in the right positions?)

Froog said...

The data I am given is almost entirely in graph form. But I am expected to write an article about this data in a few thousand words of natural flowing prose. We can incorporate some of the charts - but only maybe 5 or 6 out of 20 or 30.

So, my job, in a nutshell, is to describe the charts.

'Accuracy' probably wouldn't matter either to my commissioning editor or to his clients; but it matters to me. I can't bring myself to write utter codswallop.

John said...

"Figure 1: Well this one is completely wrong, ignore it.
Figure 2: Ah, you'll see they got the decimal point mixed up here, moving on.
Figure 3: I was going to tell you what this shows but then I noticed they'd got their own definition skewed so you can forget that one too.
In conclusion, use different data."
My fee please! No, for the report AND correcting your bloody mistakes. Froog.
Heh, I'm not being facetious; I really think you should be paid twice as much!

Froog said...

I have clarification of a sort. A Chinese colleague has dug up quite a lot of material for me on the Chinese government's analysis of the urban/rural split.

The guidelines seem to get rewritten every ten years or so. And, as always seems to the case with the texts of laws and official pronouncements here, the more elaborate they get, the less clear or specific they actually are (there's at least one clause in the latest set of guidelines which, at least in the English translation, is pure gibberish).

The general idea seems to be that all significant cities and towns are regarded as 'urban', but that smaller communities - less than 20,000 population - will not usually be given that status unless a certain proportion of the population (above 10%, on the current definition) is deemed to be "non-agricultural" (which is, ahem, NOT defined).

What that seems to mean in practice is that, in the figures I'm working with, about 20% of the population of the 20,000+ 'small towns' (population roughly between 2,000 and 20,000, with the mean falling at around 8,000 or 9,000) are considered still "rural" - and, presumably, almost everything smaller than this is too.

Froog said...

Clearly, there shouldn't really be any doubt about the 'urbanity' of what are usually described as "5th Tier cities" - communities numbering in the hundreds of thousands, sometimes in the high hundreds of thousands.

My problem arose initially from my employers' utterly bizarre - and still unexplained! - decision to come up with their own tier threshold definitions which were dramatically lower than those used by anyone else.... such that the "5th Tier" became what are usually referred to in China as "large villages" (most, but not all, of which are now defined as "urban" - although the choice of classification seems to be entirely at the whim of local party bosses).

John said...

I can only speculate on what their cunning plan might have been; my actual intention on commenting is to announce that e-mail notifications are working again! Huzzah!

Froog said...

Well, that's nice. Huzzah, indeed.

I wonder if the glitch with word verification has been repaired as well?