Statistical recording of poverty.
(Statistische Erfassung von Armut.)

*(German)*Zbl 0859.62105
Angewandte Statistik und Ökonometrie. 34. Göttingen: Vandenhoeck & Ruprecht. x, 196 p. (1991).

Summary: This paper aims at showing the possibilities of statistical poverty measurement and the limiting factors that are encountered in the process. The main attention is directed to the question of how the conceivable need of information regarding poverty can be covered with statistical means.

First of all different concepts of poverty are presented which are of importance in the scientific and social discussion of poverty. Based on this a general definition of poverty is developed in the form of a verbal characterization. By rendering this definition more precise all possible variations in meaning of the concept of poverty result from it and are the basis of all further considerations. This way the possibilities and limits of statistical measurement of poverty can be analyzed more generally in theoretical and empirical research of poverty than was possible until now. In this context the degree and intensity of poverty are also introduced as fundamental aspects of the phenomenon poverty. This is followed by an exemplary survey of information which can be important with regard to poverty.

In the following possibilities of identification of poverty are presented and reflected critically. This includes theoretical considerations concerning the identification of poverty, poverty lines for temporal and regional comparisons as well as a discussion of the problematic nature of the identification of poverty including their consequences for the empirical research of poverty.

The centre of this paper contains the construction and discussion of poverty measures. Besides a formal definition of poverty, a conceptual laying of their foundations, an exemplary presentation of possible requirements for poverty measures and a discussion of types of poverty measures, a new general definition of poverty measures results on which all further considerations are based. Following this, fundamentally possible poverty measures are developed and presented in a systematic way, where a differentiation is made between measures for the degree and the intensity of poverty. With the separate measurement of the degree and the intensity of poverty many debates in poverty research – for instance about sense and nonsense of certain requirements for poverty measures – are at least partially deprived of their basis.

After presenting possible measures for the degree and the intensity of poverty there follows a detailed discussion of complex poverty measures which have been proposed in the literature. In the process it is expounded that the idea underlying these measures to include several or all aspects of poverty simultaneously, can not be realized in a sensible way. It is furthermore demonstrated that these complex measures can be interpreted as special cases of the intensity measures developed above, and that this interpretation can make them not only sensible but also useful for empirical research. In the process it becomes clear that the complex measures from the literature contain many explicit assumptions as well as implicit ones introduced by the chosen construction formulation, which need to be discussed in their interpretation. In this connection impossibility theorems and the concept of poverty dominance are dealt with, too. This is followed by an assessment how the presented measures are suitable to meet the demand for information which exists in science, politics and in general public with regard to poverty.

Finally, the consequences for empirical poverty research are shown which are caused by the limited efficiency of the apparatus available of poverty research. These considerations include problems with data which occur in statistical practice and which are extremely difficult especially in poverty research.

At the beginning of the paper the possibilities and limits of a statistical measurement of the phenomenon poverty are inquired about. To summarize the answer can be that a multitude of information which is of importance for the scientific and social discussion of poverty can be obtained with statistical methods. But the information is always full of value judgements and contains extensive simplifications, and with statistical methods alone the possible requirements for information can by no means be covered completely.

First of all different concepts of poverty are presented which are of importance in the scientific and social discussion of poverty. Based on this a general definition of poverty is developed in the form of a verbal characterization. By rendering this definition more precise all possible variations in meaning of the concept of poverty result from it and are the basis of all further considerations. This way the possibilities and limits of statistical measurement of poverty can be analyzed more generally in theoretical and empirical research of poverty than was possible until now. In this context the degree and intensity of poverty are also introduced as fundamental aspects of the phenomenon poverty. This is followed by an exemplary survey of information which can be important with regard to poverty.

In the following possibilities of identification of poverty are presented and reflected critically. This includes theoretical considerations concerning the identification of poverty, poverty lines for temporal and regional comparisons as well as a discussion of the problematic nature of the identification of poverty including their consequences for the empirical research of poverty.

The centre of this paper contains the construction and discussion of poverty measures. Besides a formal definition of poverty, a conceptual laying of their foundations, an exemplary presentation of possible requirements for poverty measures and a discussion of types of poverty measures, a new general definition of poverty measures results on which all further considerations are based. Following this, fundamentally possible poverty measures are developed and presented in a systematic way, where a differentiation is made between measures for the degree and the intensity of poverty. With the separate measurement of the degree and the intensity of poverty many debates in poverty research – for instance about sense and nonsense of certain requirements for poverty measures – are at least partially deprived of their basis.

After presenting possible measures for the degree and the intensity of poverty there follows a detailed discussion of complex poverty measures which have been proposed in the literature. In the process it is expounded that the idea underlying these measures to include several or all aspects of poverty simultaneously, can not be realized in a sensible way. It is furthermore demonstrated that these complex measures can be interpreted as special cases of the intensity measures developed above, and that this interpretation can make them not only sensible but also useful for empirical research. In the process it becomes clear that the complex measures from the literature contain many explicit assumptions as well as implicit ones introduced by the chosen construction formulation, which need to be discussed in their interpretation. In this connection impossibility theorems and the concept of poverty dominance are dealt with, too. This is followed by an assessment how the presented measures are suitable to meet the demand for information which exists in science, politics and in general public with regard to poverty.

Finally, the consequences for empirical poverty research are shown which are caused by the limited efficiency of the apparatus available of poverty research. These considerations include problems with data which occur in statistical practice and which are extremely difficult especially in poverty research.

At the beginning of the paper the possibilities and limits of a statistical measurement of the phenomenon poverty are inquired about. To summarize the answer can be that a multitude of information which is of importance for the scientific and social discussion of poverty can be obtained with statistical methods. But the information is always full of value judgements and contains extensive simplifications, and with statistical methods alone the possible requirements for information can by no means be covered completely.

##### MSC:

62P20 | Applications of statistics to economics |

91B82 | Statistical methods; economic indices and measures |

62-02 | Research exposition (monographs, survey articles) pertaining to statistics |

90-02 | Research exposition (monographs, survey articles) pertaining to operations research and mathematical programming |