Measuring regional inequality: to weight or not to weight? Spatial Economic Analysis. When estimating regional inequality, many economists use inequality indices weighted by regions’ proportion of the national population. Although this approach is widespread, its adequacy has not received attention in the regional science literature. This paper proves that such an approach is conceptually inconsistent, yielding an estimate of interpersonal inequality among the whole population of the country rather than an estimate of regional inequality. Nevertheless, as a measure of interpersonal inequality, such an estimate is very rough (even misleading) and does not always have an intuitive interpretation. Moreover, population-weighted inequality indices do not meet the requirements for an adequate measure of inequality.
This article reviews domestic and foreign studies that empirically analyze inter-regional income inequality in Russia. The studies are grouped according to statistical approach applied (cross-section approach, time-series approach, and distribution dynamics approach). The adequacy of techniques used and data analyzed is discussed. The issue of relationship between examining inter-regional income inequality and policy implications is considered as well.
Lacking data on price levels across locations (countries, national regions, etc.) for cross-space comparisons, researchers resort to local consumer price indexes (CPIs) over time to evaluate these levels. This approach unfortunately fails to specify, even generally, the exactness of such proxies. Worse, the method is silent on whether the results are consistent, at least qualitatively, with those obtained using actual price levels. This paper aims to find an answer empirically, using data across Russian regions. Through comparison of CPI-proxied price levels with direct evaluations of regional price levels (i.e. Surinov spatial price indexes and the costs of a purchasing power basket), biases that distort the qualitative pattern of inter-regional differences are identified. Cross-region distributions for real income (calculated with CPI-proxied and directly evaluated price levels) for several points in time are estimated and compared. The CPI-induced biases are found to generally overstate inter-regional disparities.