Forecasting unemployment rates is a paramount problem for society. It is in direct relationship with the global economy, further complicating an already complex problem. Therefore, traditional techniques cannot provide the required performance in solving such problems. A possible solution lies in the field of artificial intelligence (AI). The unemployment rate can be formulated as a time-series prediction task and appropriate emerging techniques applied. This work explores the potential or reservoir computing based techniques, specifically the use of echo state networks (ESN) in order to account for the sensitive relations between various economic factors and produced more accurate forecasts. To maximize the performance of the ESN, an modified hybrid version of the red fox optimization (RFO) algorithm is applied for hyperparameter optimization. Rigorous testing indicates the promising results of the proposed technique, which is validated against state-of-the-art metaheuristic algorithms. The best performing models demonstrate a mean square error (MSE) as low as .006967 suggesting promising potential.