Government expenditures may tend to defray to increasing social needs with economic growth and development. This study focused on the effects of government expenditure on unemployment and conducted empirical research on the possible positive or negative effects of government expenditure on unemployment. This study employed the asymmetric causality and Kernel-based Regulated least squares (KRLS) method with machine learning to examine the effect of government expenditure on the unemployment rate for two Baltic countries (Estonia and Latvia) for annual time frequency data from 1995 to 2019. Asymmetric causality findings show that increase government final government consumption has a decreasing impact on unemployment in the Estonia. In addition to this, no enough evidence for Abrams Curve from Estonia and Latvia. Additionally, the government’s final consumption and expenditure effects have different signs at different quantiles. KRLS quantile coefficient findings point out that V-shape and inverse V-shape curve instead of the Abrams Curve for our sample.