بررسی فقر چندبعدیِ مناطق شهری ایران در دوره 1383 الی 1398: به‌کارگیری وزن‌های حاصل از تحلیل تناظر چندگانه در روش آلکایر-فوستر

Introduction: The extremely poor population (less than $ 1.9 daily income) has declined in the last three decades (World Bank, 2020), but more than 600 million people still suffer from severe poverty. Crises like the outbreak of the Covid-19 make the poverty even worse. Iran, as a middle-income country, has experienced a reduction in poverty in recent decades, but economic crises such as the stagflation in 2018 have increased the number of the poor.
The importance of poverty (and, of course, inequality) has brought the issue into the focus of politics and science, creating an extensive literature. To understand the nature and determinants of poverty and the ways to deal with it, many theories are introduced to pave the way for policy-making. Measuring poverty and drawing its map are the first step in the study and policy-making of it. Measuring poverty is impossible without defining it. Since the nineteenth century, when empirical research in the field of poverty began in England, the definition has undergone significant changes. Poverty was initially defined as an income below a poverty line, but, in the 1980s, the concept of multidimensional poverty was introduced and used as an alternative to income poverty. Along with conceptual developments, technical advances in the measurement of poverty have been made in both income and multidimensional poverty areas. In recent decades, these issues have spread in Iran, and various approaches have been used to measure poverty. In this regard, the present study measures multidimensional poverty in the urban areas of Iran.
Methodology: In this study, the Alkire-Foster (AF) method is used to measure multidimensional poverty. By this method, the identification stage is done in two steps, determining the level of deprivation and identifying the poor. Each of these steps has its own cutoff, which is why it is called the "dual cutoff method". The aggregation stage is done with special indices of this method. Calculating the indices requires three parameters including the dimensional deprivation cutoff vector, the indicators weight vector, and the poverty threshold. The weight of the indicators can be measured in four ways: 1) equal weight, 2) the views of experts or policy makers 3) participatory approach according to the priorities of the community and 4) statistical methods. In this study, the statistical method used for weighting is Multiple Corresponding Analysis (MCA).
MCA can be categorized under factorial methods that are formed around the concept of inertia and variance. The factorial approach has three important subsets: principal component analysis (PCA), factor analysis (FA), and MCA. As an intuitive description, PCA can be thought of as fitting an ellipsoid to a mass of n points in the d dimensional space, as the obtained elliptical diameters are the new axes. These axes are called the "principal component." Despite the advantages, the PCA method has its limitations. Among other things, this method is designed for numerical variables with the same scale. For ordinal categorical variables, the MCA method was developed as an extension of the Correspondence Analysis (CA) method.
Results and Discussion: As many as 13 indicators are selected in the four dimensions of "economic welfare", "housing", "health", and "education" with the data of the Household Expenditure and Income Survey in Iranian urban areas from 2004 to 2019. The calculation of the indicator weights by MCA is the first step. One of the factors affecting the weight of an indicator is the diffusion of that trait between households. Indicators with very low deprivation take higher weights. This feature is called the "prevalence principle". To implement the AF method, the poverty line corresponding to the headcount ratio of 30% is considered, which is equal to 0.209. Using the weights of households (i.e., the number of the households), average deprivation share across the poor and the adjusted headcount ratio in the whole sample are found to be 34.12 and 10.60 percent, respectively. After the calculation of the annual trend of the multidimensional poverty index, it emerges that this trend is downward and convex overall, but, in the last two years of the period, the headcount and the adjusted headcount ratio increase. Robustness analysis and statistical inference confirm these results. The study of regional poverty shows that Mazandaran and Sistan and Baluchestan have the lowest and the highest headcount and adjusted headcount ratios, respectively.
To compare the effects of the weighting method on the results, the poverty indices are calculated in terms of MCA, equal, and nested weights with a poverty line of k = 1/3. The headcount indices of these three methods are 0.13, 0.20, and 0.29 percent, respectively. Also, the weights obtained from MCA achieve lower indices. The analysis of the contribution of the indicators in the adjusted headcount ratio of the whole sample indicates that "the education level of the household head", "communication facilities", and "income poverty" have the highest contribution.
Conclusion: The downward and convex trend of annual indices means that, over time, poverty reduction occurs at a slower slope and, finally in the last two years, poverty indices have an upward slope. Of course, this coincides with the stagflation in the Iranian economy. The examination of the deprivation trend in the indicators also shows that the indicators of "economic poverty", "calorie adequacy", "household literacy index", and "the education level of the household head" are upward. Meanwhile, the contribution of the first two indicators in terms of the adjusted headcount index in the period under study is increasing. In other words, in the last two years of this period, the indicators of both multidimensional and income poverty increased, which requires policymakers to pay attention to this issue. Specific policies should also be developed and implemented for areas and groups that are more impoverished.