Dynamic factor models with clustered loadings: Forecasting education flows using unemployment data

We propose a dynamic factor model which we use to analyze the relationship between education participation and national unemployment, as well forecast number of students across many different types education. By clustering loadings associated with macroeconomic factor, can measure what extent exhibit similarities in their cycles. To utilize feature that unemployment data is available for longer time period than our detailed panel data, two-step procedure. First, consider score-driven filters conditional expectation rate. Second, multivariate regress on further apply k-means method estimate clustered loading matrix. In Monte Carlo study, performance proposed procedure its ability accurately capture clusters preserve or enhance forecasting accuracy. For high-dimensional, nation-wide set from Netherlands, empirically investigate impact rate choices over time. Our analysis confirms part-time covaries more strongly those full-time