Latent Dimensions of Deprivation and Child Poverty Rates in Nordic Households : An Exploratory Factor Analysis and Panel Regression

Sammanfattning

This thesis examines whether latent dimensions of deprivation exist among Nordic households without dependent children, by performing Exploratory Factor Analysis (EFA) on eight EU-SILC indicators of material and housing deprivation. Motivated by the observation that opposing claims about child poverty can both be statistically valid depending on the chosen measurement method, it is further examined whether the identified latent factors are associated with the At-Risk-Of-Poverty (AROP) rate in Nordic households with dependent children. The data cover all Nordic countries over the period 2003–2025, yielding 460 observations across four different household types without dependent children. The suitability of the data is confirmed through Mardia tests, Bartlett's test of sphericity and the KMO measure, before applying Iterated Principal Axis Factoring (IPA) with direct oblimin rotation to the Spearman correlation matrix. The estimated Bartlett factor scores are subsequently used as predictors in a fixed effects panel regression, with the AROP rate in households with dependent children as the dependent variable. Two latent factors are identified, named Financial Deprivation and Secondary Deprivation, with a strong factor correlation of 0.754. The combination of subjective and objective indicators under Financial Deprivation supports the validity of subjective measures in poverty research. In the subsequent panel regression, Financial Deprivation is statistically significant at the 5% level in both the linear and nonlinear model specifications, whilst Secondary Deprivation does not reach statistical significance at the 5% level in either specification. The panel regression results confirm that Financial Deprivation carries meaningful information about AROP rates in Nordic households with dependent children. The linear model outperforms the nonlinear model in the out-of-sample forecast, with household composition explaining much of the in-sample variation. These findings suggest that child poverty rates in Nordic countries are at least partly reflected by broader dimensions of household deprivation, rather than requiring entirely child-specific indicators.