AN AUTOREGRESSIVE INTEGRATED MOVING AVERAGE MODELING OF CROP PRODUCTION INDEX IN NIGERIA

Abstract

In univariate time series econometrics model, forecasting is an important tool for assessing

the performances of any single-variable time series such as the Crop Production Index (CPI).

This study therefore, forecast the expected or future values of the CPI series in Nigeria using

Box-Jenkins (1976) methodology. Pre-tests of the annual CPI series extracted from the World

Governance Index spanning 1961 to 2018 (58 years) confirmed that the CPI was a difference

stationary series of order one {I(1)}.The CPI data set was divided into train and test sets. The

train set, 80% of the CPI series which is approximately 46 years covering 1961 to 2006 was

used to develop the model. ARIMA (1,10), ARIMA (1,1,2) and ARIMA (1,1,1) models are

suggested and all were used on the test data covering 2007 to 2018. ARIMA(1, 1, 0) was found

to be the best among the competing models under model identification, parameter estimation,

diagnostic checking and forecasting evaluation of the test data Using RMSE, MAE and

MAPE performance indicator indices. Post-estimation test using a simple residual

correlogram further disclosed that the residual obtained from the fitted model was white

noise (i.e. all spikes of the plot are within the 95% confidence bounds). Lastly, The Out of

sample forecast of the CPI using ARIMA(1,1,0) for the next 12 years (2019 to 2030) shows an

upward trend with a constant growth of 1% to 2% annually. It is therefore recommended that

efforts should be geared towards improving agricultural productivity by all stake holders in

Nigeria to overcome the challenges of food security by the year 2030