Improving Unemployment Rate Forecasts Using Survey Data

This paper investigates whether forecasts of the Swedish unemployment rate can be improved by using business and household survey data. We conduct an out-of-sample forecast exercise in which the performance of a Bayesian VAR model with only macroeconomic data is compared to that when the model also includes survey data. Results show that the forecasting performance at short horizons can be improved. The im-provement is largest when forward-looking data from the manufacturing industry is employed.