Modeling and Forecasting Unemployment Rate in Tanzania: An ARIMA Approach

Abstract

Purpose: This study aims to develop a reliable forecasting approach for Tanzania's unemployment rate and provide policymakers with an effective tool for decision-making. Unemployment forecasting is vital for informed policymaking, particularly in countries like Tanzania. Methodology: This study employs a quantitative research design and adopts Box Jenkin's methodology and the ARIMA (AutoRegressive Integrated Moving Average) model for unemployment forecasting in Tanzania. The entire available dataset for the specified period is utilized, employing a non-probability sampling technique. Diagnostic tests, including ACF (AutoCorrelation Function), PACF (Partial AutoCorrelation Function), and unit root analysis, are conducted to guide the optimal model selection. Differencing addresses non-stationarity in the time series data by removing trend and seasonality effects. The optimal model selection is based on criteria such as AIC (Akaike Information Criterion), Schwartz, and Hannan-Quinn. Findings: The study finds that the ARIMA (3,1,4) model demonstrates superior performance in forecasting the unemployment rate in Tanzania. Diagnostic checks validate the adequacy of the model, revealing white noise residuals. The forecasts indicate a consistent downward trend in unemployment rates over the next nine years, suggesting potential labour market improvements in Tanzania. These findings enhance our understanding of Tanzania's unemployment dynamics and provide valuable insights for policymakers. Originality/Value: The study lies in its application of Box Jenkin's methodology and the ARIMA model to unemployment forecasting in Tanzania. By utilizing the entire available dataset and employing diagnostic tests for model selection, the study enhances the reliability of the forecasting approach. The study offers policymakers an informed decision-making tool by providing accurate forecasts and capturing underlying trends.