This paper develops new methodology for normative-statistic estimates of poor-specific purchasing power parity (PPP). We show that the only way of poor-specific PPPs constructing consistently with poverty measurements is determining them as ratios of poverty cut-offs within regions or countries to a benchmark poverty line. The key idea of normative-statistic definition of poverty cut-off consists in fixing a controlling welfare indicator, having unambiguous normative definition, and estimating aggregated income or expenditure level under which expected level of controlling indicator coincides with its fixed level. Usually the controlling indicator chooses as total expenditures for food consumption and its fixed level chooses as the cost of minimal food basket. But we introduce additional correction of normative-statistic poverty cut-offs by analysis of consumer basic needs in non-food goods and services. Normative-statistic method of determining poverty line works out with distributions of income and expenditure estimated on the base of HBS data. We introduce new method for exact estimating normative-statistic poverty cut-offs, using balanced kernel estimates of distribution density functions for a collection of welfare indicators. In this paper we approve our methodology on the full set of Russian regions. This evidence is instructive and suggestive, because socioeconomic differentiation of Russian regions is comparable with the world differentiation of countries. But proposed methodology is applicable to any number of countries and gives estimates of poor-specific purchasing power parities stable under addition of new countries.