politics : culture : economics

Damned statistics: poverty rates in Turkey’s regions

In Economy, Turkey on August 7, 2010 at 9:23 am

Turkey’s Income and Living Conditions Survey for 2008 was released at the end of July. As is always the case with these releases, and as the table at the foot of this article makes clear, it’s worth having a rummage around in the data to see what patterns may have been overlooked.

When the survey was published, considerable attention was paid to TurkStat’s headline announcement that the ratio  between the incomes of the richest and poorest 20 per cent of society was unchanged at just above 8:1 between 2007 and 2008. A bit of international context helps to put this figure in perspective. Among OECD members, only Mexico has a less equal distribution of income than Turkey. In 2005, the ratio between the income of the top and bottom 10 per cent averaged 9:1 across the OECD. In Turkey the corresponding figure was 17:1.

Given this wide gap between high-income and low-income individuals in Turkey, and given also the massive variation in levels of economic activity in different parts of the country, one would expect to see sharp regional differences in poverty levels. In particular, one would expect to see much higher-than-average rates of poverty in the under-developed eastern and southeastern regions.

At first glance, therefore, it’s surprising to see that TurkStat’s assessment provides such a uniform picture of poverty across the country. (Use this link to see the relevant excel table from TurkStat’s release.) Out of 12 regions, ten are listed as having a poverty rate in a narrow range from 9.5 per cent to 13.7 per cent. And somewhat surreally, the poverty rate in the southeastern Anatolia region, at 12.7 per cent, is less than three percentage points higher than the rate in Istanbul.

The reason for this pattern is the way that TurkStat is defining poverty in the regions. Their basic definition is uncontroversial—an individual is said to be in poverty if their after-tax income is less than half of the median income. On a national basis, that gives a 2008 poverty threshold slightly higher than 3,000TL and a poverty rate of 16.7 per cent (ie, 16.7 per cent of people in Turkey had an after-tax income lower than the threshold figure).

But the regional TurkStat figures in the table linked to above don’t assess regional incomes against this national poverty threshold. Instead, they create a separate poverty threshold for each region based on a separate calculation of the median income of that region’s inhabitants. This move inevitably flattens out variations in the rate of poverty. An individual on a low income in the southeast of the country is obviously going to appear less poor if they are compared against the incomes of their regional peers rather than against the higher national average.

The table below highlights the extent of the discrepancy between TurkStat’s regionally ‘tailored’ measures of poverty and the measures that result when regional incomes are compared against the nationwide median income.

The first and second columns list TurkStat’s regional and national poverty thresholds, in Turkish liras. The third column lists the poverty rate in each region on the basis of the regional poverty threshold. The fourth column lists the poverty rate in each region on the basis of the national threshold. As far as I’m aware these fourth-column figures aren’t provided in TurkStat’s dataset. I’ve calculated them using a number of other figures that are provided there. They paint a very different picture of the rates of poverty that exist in certain parts of the country.

Regional threshold (TL) National threshold (TL) Poverty rate (regional basis) Poverty rate (national basis)
İstanbul 4 574 3 146 9.9 3.2
West Marmara 3 369 3 146 12.9 11.6
Aegean 3 540 3146 15.7 11.4
East Marmara 3 992 3 146 9.5 4.9
West Anatolia 3 596 3 146 12.7 9.0
Mediterrannean 2 597 3 146 10.0 16.6
Central Anatolia 2 867 3 146 12.7 16.3
West Black Sea 2 795 3 146 11.9 16.8
East Black Sea 3 318 3 146 13.7 11.6
North East Anatolia 2 189 3 146 17.7 34.0
Central East Anatolia 1 838 3 146 9.3 36.8
South East Anatolia 1 550 3 146 12.7 47.9

Source: TurkStat, Income and Living Conditions Survey, 2008

Add to: Facebook | Digg | Del.icio.us | Stumbleupon | Reddit | Blinklist | Twitter | Technorati | Yahoo Buzz | Newsvine

  1. Hi there,

    Just to follow upon the discussion we started on twitter @livinginturkey…

    I had a look at the turkstat website and found the data you actually wanted: Regional Areas Vs National Average. Here you see poverty level of Istanbul as 3.5% and poverty level for South East as 30.4% as expected.

    Please have a look at the data called “Number Of Poors And Regional Poverty Rates By Relative Poverty Thresholds (Calculated For Turkey) Based On Income ” http://www.turkstat.gov.tr/PreIstatistikTablo.do?istab_id=1287 on page http://www.turkstat.gov.tr/VeriBilgi.do?tb_id=24&ust_id=7

    While your figure for Istanbul is close to the TurkStat one you have a 47.9% Vs Turkstat 30.4% for the South East.

    My guess is that you didn’t consider household and did not perform an equivalisation (Cf. http://en.wikipedia.org/wiki/Equivalisation and http://www.poverty.org.uk/summary/income%20intro.shtml ) Here the idea is you need to adjust for household’s size while taking into account economies of scale.

    So anyway the figures you wanted were actually available at Turksat.

    FYI also I had a look in UK and France:
    – UK:http://www.poverty.org.uk/02/index.shtml as you can see London has got the highest percentage of poverty in the UK which tends to indicate the UK too on its main stats are calculating a regional threshold too, otherwise we should have a much much lower figure Vs UK median. Did not see regional comparisons Vs national data
    – France: http://www.insee.fr/fr/themes/tableau.asp?reg_id=20&ref_id=revop004 I found stats for the Area around and including Paris (Il de France) and they give both according to national (3.7%) and regional (8.6%) but could not find any other region.

    So in the end it seems that the Turkish stats are quite comprehensive and contain comparisons Vs both National and Regional median

    Finally it seems (as described here http://www.poverty.org.uk/summary/income%20intro.shtml) that most researchers put more weight on a relative scale to measuring poverty. SO basically regional threshold seems more meaningful there as Threshold should be drawn in the context of the household considered: socially that threshold is meaningful Vs most households in the region. So in SE only 12.7% of household wd feel “poor” and not 30.4% as the local circumstances are very different from national average.

    Anyway I do hope that most of this ranting sort of makes sense, getting very late! Sorry to bother you on that, I need to say I really appreciate all your articles, i just felt that for this one some extra perspective was needed. Take care!

    • Rob,

      Thanks for the comment. I had seen those other TurkStat figures that you link to. They’re actually doing something different from what you think they’re doing. They don’t provide a poverty rate for each region (ie, the percentage of the regional population below a poverty line). Rather, they tell you what proportion of the countrywide number of people below the (countrywide) poverty line lives in each region. (That’s why all the regional figures sum to 100 per cent; and it’s why the figure for Turkey in cell N7 is 100 per cent – that’s not telling us that there’s a 100 per cent poverty rate in Turkey, it’s just saying that all of Turkey’s impoverished people live in Turkey.)

      The figures I was using were the equivalised ones, so household-size factors would have been taken into account. So I’m going to stick to my guns in terms of 47.9 per cent as being the proportion of people in poverty in the SE Anatolia region when a poverty threshold of 50 per cent of (countrywide) median income is used.

      As I mentioned before in the twitter thread, I’m not suggesting that intra-regional comparisons don’t have a role. It’s just that a comparison of regions against a countrywide baseline provides a very useful picture of relative conditions in different areas. Your underlying point is that poverty should be measured according to local norms, to reflect who “feels” poor relative to others in their geographical vicinity. The danger of that approach is that it limits the usefulness of poverty statistics as a public policy tool for identifying and countering income inequality across the country. And in Turkey there are very wide inequalities between different parts of the country.

      • Autant pour moi! as we say (roughly “I stand corrected!” ;))…I could not find a hole big enough to bury my body in it so there you go…I did mention it was late and you are perfectly right the figures I wrongly quoted are of course just distribution of poors in Turkey.

        From the many tables we have we can see that 30% of Turkish Poor according to the 50% median definition live in the South East. In numbers it is around 3.3 millions.

        Now South East seems to have a population of around 6.8 millions so with straight logic around 48% of people in SE live in poverty as defined nationally. As you found out!

        So I stand corrected but I learnt in the process of looking at those numbers, for example the definition you used is the standard one in OECD and Europe. For policy making this is the measure of choice. Although the basket of goods approach is also used by the US for example.

        I also just meant that someone below the 50% threshold in Istanbul and Diyarbakir and same income would not feel as poor in Diyarbakir for basket of goods and social reasons. But of course not a reason not to have policy to develop SE. Especially developing SE is part (just part) of a solution to long lasting Kurdish problem.

        Thanks for bothering with my ranting. Have a good sunday!

  2. […] Survey for 2008 was released at the end of July. As is always the case with these releases, and as the table at the foot of this article makes clear, it’s worth having a rummage around in the data to see what patterns may have been […]

  3. […] in the east In Economy, Society, Turkey on August 15, 2010 at 9:26 pm I followed up my recent blog post about TurkStat’s estimates of regional poverty rates with a piece in Today’s Zaman today. The core focus remains the same: the skewing of […]

  4. […] This is true in financial terms, particularly in the country’s south and southeast, where poverty levels are scandalously high. It is true in cultural terms, with particular problems again in the Kurdish south and southeast. […]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s