Although the figures given during the election campaigns show a progress in the reduction of poverty, are these enough to assert that the poverty has diminished in Turkey? What do different thresholds mean in the poverty measuring? What is the best poverty line measure to compare a population within time series or with other populations? Is it adequate to solely use the number of people under a certain threshold to claim the poverty has reduced? In this analysis, I examine these questions not only to give some answers, but also to provide a brief yet comprehensive account of poverty measurement. Thus, this brief account of poverty analysis could be employed by the political parties and increasingly active civil society actors, such as NGO’s, in assessing the conditions with an environment, in which the data are abundant to be utilized.
Analysis of the data on poverty is concerned with people who have income or consumption that is less than a determined threshold, which is called as poverty line. In Turkey, the official data on poverty based on poverty lines are prepared by the Turkish Statistical Institute through “Household Budget Surveys,” which has been regularly published since 2002. Poverty analysis consists of two steps: defining the standards of the poverty line and analyzing and comparing the results of a set of measures. The standards and the measures that are employed in the analysis bear considerable significance since they have different implications and various qualities.
There are three ways, by which the threshold is defined: absolute, relative and subjective standards (Chart 1). Absolute poverty line is conducted by determination of a basket of basic needs, which might include food, non-food, and fundamental services that an individual needs in the daily base. The expenditure of this basket is indexed to the market prices and used as the standard of the poverty line. Hence, an objective standards of poverty is obtained in order to measure the poverty with regard to many dimensions. However, since conducting an objective basket is hard to achieve that corresponds basic needs of people from different socio-economic backgrounds and countries, the some monetary amounts, such as $1, $1.25, and $2, are employed with regard to Purchasing Power Parity (PPP). Recently, World Bank has raised the global poverty lines to $1.90 and $3.10 respectively, whereas Turkish Statistical Institute (TurkStat) sets the threshold as $2.15 and $4.30 based on the current PPP values in TL ($1 threshold has not been calculated since 2006 due to the assumption that there is no people under it anymore). Even though employing certain predetermined monetary amounts facilitates the comparison, it might fail capturing societies’ varying perceptions of need with regard to unessential or non-material needs, such as those that are accepted necessary for social and psychological interaction. In order to tackle this problem, relative standard is employed in defining the poverty as fixed fraction of median income or consumption in the population. In other words, number of people that are under poverty line is determined relative to the median income or consumption. Thus, it can be concluded that there is strong correlation between relative standards and inequality. Indeed, based on relative poverty, the poverty rates of some high income countries could be turned out higher than the low income yet more equal countries in the World. The EU sets its official relative poverty line 60 percent of equivalised disposable national median income whereas TurkStat sets it 50% even though it provides public with data based on 40, 50, 60 and 70 percent relative poverty standards.
Although it is intuitive to compare poverty values of different populations based on relative poverty standards, it might cause some problems in some cases of times series, in which the relative poverty rate might not change since whole distribution, thus median income, shifts at the same time. Nevertheless, this problem might be ignored in the cases that whole distribution gets wealthier due to the fact as getting wealthier, the society’s perception of poverty might change as well. However, in the scenarios that whole distribution become poorer than before, it might fail capturing the severity of the poverty that deteriorated. Especially, many indicators based on relative poverty lines have misled the people by understating the reality. Furthermore, unless the national median income is adopted rather than a single average income in the comparison among different populations, the relative poverty rates of various countries would be correlated with the income of that country. While these two definitions of poverty lines are widely employed in measuring poverty, subjective standards are less utilized because it is conducted through personal interviews and surveys; therefore, it is probable the standards would be biased by psychological and ideological standpoints that are hard to be reconciled with objectivity.
Once the appropriate standard for poverty line is employed for the measurements, the data could be interpreted through employing different means of measures. Some of the indicators that are widely employed are the Headcount Ratio, the Income Gap Ratio, the Poverty Gap, the Squared Poverty Gap and the Sen Index. However, not all these poverty indicators have the same implications due to the fact that they vary in terms of qualities that they provide as insight. Indeed, to comprehend the intuitions that each measure provides is crucial in order to interpret the data on poverty gathered based on appropriate poverty line. Three significant dimensions, about which a perfect poverty measure could be intuitive, are incidence, intensity, and inequality, namely “Three I’s of Poverty.” While the incidence means the measure is able to show the number of person under the poverty line, the intensity implies the severity of poverty, and the inequality means whether the measure reveals the degree of inequality among the poor.
Thus, there are some principles –axioms- that are used to qualify the measures in order to understand their limitations. Some of these essential principles are focus axiom, population principle, monotonicity, and Pigou-Dalton transfer principle. While the first axiom implies the measurement is concerned with only the proportion of people that lives under the poverty line, the second axiom is the quality that the measure does not change when the population is replicated. All of the indicators that are mentioned above satisfy these two axioms. Monotonicity and Pigou-Dalton transfer principle, on the other hand, are hard to be qualified since they require more intricate structures as not only poverty but also inequality measures. Monotonicity axiom means that inequality that the indicator reveals must increase (decrease) when the income of any poor person decreases (increases). Furthermore, the Pigou-Dalton transfer principle implies that the inequality must increase (decrease) when there is an income transfer from a relatively resource-poor (rich) person to a resource-rich (poor) individual.
The headcount ratio, which is the proportion of the poor in the whole population, is the most commonly equipped poverty measure since it is easy to conduct and to be compared. If the headcount ratio is 50%, it means 50% of the population is under the poverty line. Nonetheless, it satisfies only the focus and population axioms and it shows only incidence dimension of poverty. Hence, it might be not only less informative but also misleading. For instance, when a transfer from the least poor to individuals, are just below the poverty line, makes them to pass the threshold, the headcount ratio cannot reflect the poor who are worse-off due to transfer (Figure 1). Thus, another measures should be examined alongside with the headcount ratio.
In addition, Income Gap Ratio is the average difference between the poverty line and the average income of the poor. If the income gap ratio of a population is 20%, it implies the average income of the poor equals to the 80% of the poverty line. Income gap ratio is able to show the intensity of the poverty whereas it lacks intuition about incidence and inequality. Furthermore, like headcount ratio, it qualifies for the focus and population axioms while it violates the monotonicity and transfer principles. Indeed, for example, when individuals that are just below the poverty line are transferred money that enables them to surpass the threshold, the poverty measure should show less poverty whereas income gap ratio reveals more poverty since the average income of the poor decreases by the upward movement of the top income earners or consumer of the poor over the threshold (Figure 2).
By combining the headcount ratio and income gap ratio, the Poverty Gap is obtained that is able to show the both the incidence and intensity of the poverty at the same time. It is a useful measure that satisfies monotonicity unlike headcount ratio and income gap ratio. However, it cannot fulfill the requirement of the transfer principle since it cannot reflect a transfer that takes place with the poor that does not make any poor to go beyond the poverty line. In other words, it cannot depict the changes in the inequality with the poor (Figure 3).
To measure inequality, besides incidence and intensity among the poor, the Squared Poverty Gap is utilized, which is able to reflect the results of transfer of income within the population of the poor that is also sensitive to the changes in the inequality. However, it is hard to interpret without considering other indicators, which makes it hard to be read. The Sen Index is the prepared by Amartya Sen in 1976 in order to conduct a measure that satisfies all axioms, including the transfer principle, by means of addition of Gini Index of the poor. Furthermore, it could reflect all three I’s of the poverty. One limitation of the Sen Index could be stated as the fact that it cannot be decomposed. However, it is hard to obtain the Sen Index data on a regular base, which makes it hard to be employed in updated research.
The Case of Turkey [vii]
In this section of the analysis, I will apply the theoretical part that I mentioned in the preceding parts to the case of Turkey to give some answers the questions that are posed in the introduction.
In the election campaign of the Justice and Development Party, the absolute poverty standards were employed in terms of $1, $2.15, and $4.3 values. Furthermore, based on the results of headcount ratio, it is concluded that the poverty is reduced considerably. The figures on the Chart-2 are in accordance with the program’s claim. Nevertheless, as mentioned above, the absolute standards fail at capturing the non-extreme poverty, including the needs that the society perceives necessary for the individuals that are bounded to its norms. Moreover, Turkey experienced considerable increase in the GDP per capita from $8667 in 2002 to $19610 in 2014 in PPP. [viii] Thus, using only absolute standards would be far from sufficient to grasp the poverty in a fast developing country like Turkey. Indeed, as aforementioned in the previous sections, relative poverty would fit for the developed and fast developing countries. Although Turkish income distribution experienced a considerable shift, it might be ignored since it was a positive shift that resulted in the increase in the incomes of whole distribution. Thus, when the Chart-3 is examined, which indicates Turkey’s income-based headcount ratio (poverty rate) with relative standards of 60% as the EU uses, the number of people under the poverty line decreased from 24.5% in 2006 to 20.9% in 2014, which could be interpreted as modest decrease with regard to the considerable decrease in the absolute poverty line although 20.9% constitutes enormous number of the poor. In addition, since relative poverty line is more correlated with inequality measures, one might conclude that the huge increase in the GDP per capita has relatively less impact on the progress toward equality.
So far, both in this part and the party campaign program, the incidence dimension of poverty is concerned, which is not adequate to examine the poverty alone. The intensity and the inequality dimensions of poverty, on the other hand, have not been touched upon. The intensity of poverty, which is grasped through income gap ratio (TurkStat refers it as the poverty gap due to methodological differences), varies between 20-30% throughout time, which means average percentage of the poor is the 70-80% of the relative poverty line (Chart 3). With regard to the fact that the extreme poverty has reduced considerably, it could be concluded that the intensity of the poverty is not severe despite the fact that there has been almost no considerable increase in the average income relative to the respective poverty lines (here, it should be considered that since the inco me gap ratio decreased to a little degree, the problem that emerges as increasing income gap ratio when the headcount ratio increases could be disregarded). Here, an important issue that should be pointed is that TurkStat has been preparing its data on intensity of poverty not based on average income of the poor but the median income of the poor, which might result in quite different dataset. Inequality among the poor, however, is very hard to measure with the data provided by the TurkStat. That is because it is possible to reach the results of Squared Poverty Gap yet hard to interpret whereas it is too hard to conduct Sen Index since it requires GINI of the people only under the poverty line, which is not given by the TurkStat. Thus, it could be concluded that in Turkey, the incidence dimension of poverty varies considerably based on the standards employed.
Furthermore, the intensity of the poverty and the inequality among the poor are largely ignored, and many political parties and actors make haste to reach conclusion about the poverty. Lastly, we should consider the fact that the millions of Syrians that have taken refuge in Turkey are not included into this dataset, which implies that what three dimensions of poverty indicate might be quite different than we assume.
Alper Şükrü Gencer
[i] Turkish Statistical Institute (TurkStat). (n.d.). Statistics. Retrieved June 17, 2016, from http://www.turkstat.gov.tr/
[ii] OECD (n.d.). Retrieved June 23, 2016, from https://data.oecd.org/gdp/gross-domestic-product-gdp.htm
[iii] Pinotti, P. (2013). Tools for Income Distribution Analysis [Powerpoint Slides, PDF].
[iv] World Bank. (n.d.). Poverty & Equity. Retrieved June 17, 2016, from http://povertydata.worldbank.org/poverty/country/TUR
[v] Justice and Development Party (n.d.). Election Statement 2015. Retrieved June 23, 2016, from http://www.akparti.org.tr/upload/documents/2-insani-kalkinma.pdf
[vi] World Bank. (n.d.). Measuring and Analyzing Poverty. Retrieved June 17, 2016, from http://web.worldbank.org/WBSITE/EXTERNAL/TOPICS/EXTPOVERTY/EXTPA/0,,contentMDK:22405907~menuPK:6626650~pagePK:148956~piPK:216618~theSitePK:430367,00.html
[vii] The Poverty Site. (n.d.). Poverty and Inequality in the European Union. Retrieved June 17, 2016, from http://www.poverty.org.uk/summary/eapn.shtml
[vii] Pinotti, P. (2013). Tools for Income Distribution Analysis [Powerpoint Slides, PDF].