Seasonal Autoregressive Integrated Moving Average model for tax revenue forecast in Kenya

Fredrick Kyalo Samuel1 and Titus Kithanze Kibua 2

1Department of Mathematics and Actuarial Science, Kenyatta University, Nairobi, Kenya
P.O. Box 8747-00300, Nairobi, Kenya
E-mail: kyaloo2004@yahoo.com
2Department of Mathematics and Actuarial Science, Kenyatta University, Nairobi, Kenya
P.O. Box 43936-00100, Nairobi, Kenya
E-mail: tkibua@yahoo.com

ABSTRACT
Modelling and forecasting tax revenue is desirable in an economy for long-term projections and proper fiscal planning. This study sought to establish Seasonal Autoregressive Integrated Moving Average (SARIMA) model for tax revenue forecast in Kenya. SARIMA (2,0,0)(2,0,0)12 was identified as an appropriate model based least Akaike Information Criteria (AIC) value. The model passed residual normality test and the model residuals followed a white noise process. The predictive ability tests of Root Mean square (RMSE) and Mean Absolute Error (MAE) revealed that SARIMA (2,0,0)(2,0,0)12 was accurate, consistent and appropriate for forecasting tax revenue. The Monte Carlo simulations of tax revenue using SARIMA (2,0,0)(2,0,0)12 model produced similar plots for original and simulated time series models. The tax revenue forecasts will exhibit the similar patterns with no significant growth in the next five years. The study recommended application of SARIMA (2,0,0)(2,0,0)12 model in tax revenue forecast in Kenya and enhancing tax revenue collections.

Key words: Tax revenue, SARIMA model and forecasting

Cite this article:
Samuel, F.K. & Kibua, T.K. (2019). Seasonal Autoregressive Integrated Moving Average model for tax revenue forecast in Kenya. European International Journal of Science and Technology, 8(6), 11-30.