Aim: According to OECD, the financial well-being score for Poles is just 9.1 out of a total of 20, whilst the Eurostat reports one of the lowest scores for Polish society. Most of the previous research primarily relied on cross-sectional data. The aim of this study is to provide insights into the income perception of Polish families and its evolution over a 15-year period, using data from a Polish national longitudinal survey. Methodology: Taking specification of the Social Diagnosis data, the author adopted a mixture of latent autoregressive models. As a result, the study accounted for the unobserved heterogeneity and provided the mean and correlation coefficient for each component of the mixture, as well as described the effect of the observed covariates. Results: The author identified groups (three mixture components) of families with a similar perception of their financial situation, showing that some families (i.e. one-parent families, multi-families) and those with more children, living in suburbs, and those professionally inactive, are in particular need of greater protection. Implications and recommendations: There is a limited number of well-rounded financial education programmes that target socio-economic groups other than children and young people in Poland. The author believes that considering subjective information about income evaluation may also help to better recognise the specific financial education needs of people in different stages of life, characterised by various socio-economic features. The main limitation of the study concerns the availability of the most recent data (new waves of the survey are not published any more), therefore future research could assess the material condition of Polish families and compare the results based on other sources of data. Originality/value: The article present a new approach to the study of the economic perception of Polish families, and deals with the problem of the unobserved heterogeneity. This approach also allows to account for observable (socio-economic) characteristics and survey weights. Moreover, the author compared the results for the presented approach with the other latent variable model techniques.
Ewa Genge (Thu,) studied this question.