Application Of Auto-Regression Integrated Moving Avaerage Model To Forecast Tea Prices At Mombasa Tea Auction Center In Kenya

ABSTRACT

Tea production is the mainstay of many people in Kenya. The contribution of tea sector

to the Kenyan economy is enormous as it accounts for about 4% of Gross Domestic

Product (GDP). Favourable product prices act as a trigger for any commercial activity

over time and also suggest sustainability of the activity. Tea is a very significant crop

in Kenya and it is one of the major agricultural exports of the country. However, the

industry has experienced many challenges such as stiff competition from other tea producing

countries like Sri-Lanka, unfavourable climatic changes, land competition from

other crops like sugar cane just to mention a few. These challenges have caused seasonal

variation in tea pricing at the Mombasa tea auction which is the second largest black tea

auction center in the world after Colombo in Sri-Lanka. There exist plethora of studies to

justify that a careful and precise selection of Auto-regressive Integrated Moving Average

(ARIMA) model can be fitted to time series data to forecast with better accuracy the future

prices of tea auction. We have used time series ARIMA model to forecast tea pricing

at Mombasa tea auction in Kenya. The general objective of this project is to study the

trend and seasonality, and to forecast future tea prices. Monthly pricing data between

1995-2018 from International Tea Committee; African tea brokers limited was used for

model validation. To identify the best ARIMA model R statistical software was used to

analyse the tea pricing data. ARIMA (1, 1, 2) with Akaike Information criterion (AIC)

and Bayesian information criterion in table 4.1. The model was then used to predict tea

prices from 2019-2023 as shown in table 4.13.

Read Download