Kostnadsutjämningen inom Barnomsorgen i Sverige - en diskriminantanalys
BarnomsorgKostnadsutjämningDiskriminantanalysStatistikStatisticsOperations researchProgrammingActuarial mathematicsStatistikOperationsanalysProgrammeringAktuariematematikMathematics and Statistics
Sweden?s District councils and Country councils are in itself to a large extent responsible for the state financed welfare. Because of this, Sweden has a contribution- and adjustment system. The reason for this system is to create equal chance, for all District councils and Country councils, to be able to give the citizens service independently of their income and non-influential structural differences. The adjustment is built upon the so-called standard cost method, which contains four different parts. These are called income-adjustment, cost-adjustment, general state contribution and rules of incorporation. The system is also designed to separate the standard cost of District councils and Country councils. The cost-adjustment for the District councils, in which we have chosen to look more closely into, is build upon a number of different sections. We have also restricted the investigation to the childcare section. The structural cost differences for each section are calculated based upon factors that are able to reflect the structural needs- and cost-differences. Within the childcare section these factors are Tax levels, Levels of gainful employment amongst parents and Population density. The purpose in this essay is to perform a discriminant analysis of the three factors in order to test if it is possible to, in advance, predict if a District council would win or lose within the cost-adjustment system. In other words, is it possible to predict if a District council?s actual cost is bigger or smaller than its standard cost? We have also chosen to analyse four different years (2000-2003) in order to investigate if there exist similarities or differences in the results of the classification. The discriminant analysis is performed with three group variables (winners, neutrals and losers). After results from a Box test we adopt the method quadratic discriminant analyse. Generally we got poor classifications. In spite of this, it was possible to show that most of the discriminations classified better than by chance. But still, they are not acceptable. There are no differences between the years. It is not possible to predict, in advance, if a District council is a winner or a loser.