DEVELOPMENT OF AGRICULTURAL PRODUCE SALES FORECASTING SYSTEM: A CASE STUDY OF COCOA PRICES

CHAPTER ONE INTRODUCTION

1.0   Background of the Study

Over several years, cocoa bean is one of many commodities which are consumed worldwide. Cocoa bean has a wide variety of uses and become raw ingredients for many food and beverage products. One of the most popular products made from cocoabeanischocolate.Aschocolatebecomepopularandpeoplearoundtheworld enjoy chocolate in different form, the global cocoa market is also expanding.Most oftheglobalcocoabeanproductioncomesfromAfricacontinent.Alongthecocoa bean history, most of cocoa beans have been exported to Europe and USA for grindingprocessing.

Traditionalexportcropshaveplayedanimportantroleinthedevelopmentofmany African countries by generating foreign exchange earnings, government revenues, and household incomes. Dependence on a few export commodities has oftenmade these countries vulnerable to international price volatility, however, and a continued deterioration of agricultural commodities’ terms of trade has reinforced the divergence in economic development between primary-commodity-producing countriesandmanufacturingandserviceexporters(Wilson1984;UNCTAD2005).

Ghana,developingcountrythatbecameworld’ssecondlargestcocoaproducerand depend on cocoa as one of the most important agricultural export product and having a better understanding about cocoa bean prices characteristics is extremely important. Ghanaian cocoa stakeholder should pay attention to internationalcocoa

price,especiallytheproducer.Theaverageofinternationalcocoabeanpriceusually becomes a reference for Ghanaian government to determine the cocoa bean standard price. As international cocoa price is liable to change rapidly and unpredictably, it represents a price risk for the cocoa stakeholder. Therefore, the good estimation of future cocoa price becomes something important forthem.

Instability of cocoa prices creates significant risks to producers, suppliers, consumers, and other parties that are involved in the marketing and production of cocoabeans,particularlyinGhana.Inriskyconditionsandamidstpriceinstability, forecastingisveryimportantinhelpingtomakedecisions.Accuratepriceforecasts areparticularlyimportanttofacilitateefficientdecisionmakingasthereistimelag intervenesbetweenmakingdecisionsandtheactualoutputofthecommodityinthe market.

Modelling or forecasting of agricultural price series, like that of other economic time series, has traditionally been carried out either by building an econometric model or by applying techniques developed for analyzing stationary time series. Time series forecasting is a major challenge in many real world applications such asstockpriceanalysis,palmoilprices,naturalrubberprices,electricityprices,and flood forecasting (Assis et al., 2010).

It is clear that given the significance of cocoa in the modern world, the ability to provide accurate forecasts into the future price of cocoa will be of utmost importance.Moreover,therearebenefitsfromfindingtherightmodelthatforecasts the cocoa price more accurately thanothers.

In developing any time series model, parameter estimation is one of the crucial steps. Common methods of estimation include method of moment, least square estimation and maximum likelihood estimation. Method of moment on the other hand is rarely used in time series analyses because it produces poor estimates. Although it is easy, Method of Moment Estimation is not an efficient estimation method for Autoregressive Integrated Moving Average (ARIMA) model because it works for only Autoregressive models of large sizes.

1.1   Statement ofproblem

Asearlierstatedinriskyconditionsandamidstpriceinstability,forecastingisvery important in helping to make decisions. Accurate price forecasts are particularly important to facilitate efficient decision making and in any time series models parameter estimation is a crucial steps toconsider.

Parameter estimation is a discipline that provides tools for the efficient use of data for aiding in mathematically modeling of phenomena and the estimation of constants appearing in these models (Beck and Arnold, 1977). Much of parameter estimation can be related to four optimization problems:

• Criterion : the choice of the best function to optimize (minimize or maximize)
• Estimation: the choice of the best method to optimize the chosenfunction

• Design: optimal implementation of the chosen method to obtain the best parameterestimates

• Modelling: the determination of mathematical model which best describes the system from which data are measured, including a model of the error process.

The forecasting of Cocoa prices at international market will be useful for investment purpose in Ghana and parameter estimation is one of the crucial steps in developing appropriate modelling for forecasting. Maximum Likelihood is widelyusedbyscholarsforparameterestimationintimeseriesmodelbutdoesthis methodminimumforecasterror,comparingtootherparameterestimationmethods, especially Least Square Estimation, hence the need for thisstudy.

1.2   Objective of theStudy

This research is undertaken to:

1. Develop a forecasting model for monthly cocoa prices at the international market.
2. ComparetherelativeperformanceofparameterestimationswithMaximum Likelihood and Conditional Least SquaresEstimations.

• Predict a one year cocoa prices based on the optimalmodel.

1.3   Justification

The outcome of this study may inform cocoa industries about the relationship that exist between the present, past and random values of these traditional products for price forecast in economic decision. Again, researchers will also be informed on parameter estimation of maximum likelihood and conditional least square estimation as to which parameter estimation perform better in forecasting.

1.4   Methodology

ThedataforthisstudyconsistofasecondarydatafromtheCentralBankofGhana. Monthlycocoapricesoftheinternationalmarketisbeenusedforthisinvestigation.

One of the widely used univariate time series models in many research practicesis Box-Jenkins modelling. The Box-Jenkins modelling is one of the most powerful forecastingtechniquesavailableanditcanbeusedtoanalyzealmostanysetoftime series data (Christodoulos et al., 2010).

There are a number of approaches available for forecasting economic time series such as Exponential Smoothing Methods (ESM), Autoregressive models (AR), MovingAveragemodels(MA),AutoregressiveMovingAveragemodels(ARMA), Autoregressive Integrated Moving Average models (ARIMA), Seasonal Autoregressive Integrated Moving Average models (SARIMA), Autoregressive Conditional Heteroskedasticity models (ARCH),Generalized

Autoregressive Conditional Heteroskedasticity models (GARCH) as well as other representatives (including EGARCH, TGARCH, CGARCH, FIGARCH…) and Markov Switching Multi-Fractal techniques (MSMF). The approach that we will adopt includes only the historical time series of prices being forecasted, is known as univariate forecasting. ARIMA modelling is a specific subset of univariate modeling, in which a time series is expressed in terms of past values of itself (the autoregressive component) plus current and lagged values of a white noise error term (the moving average component). This paper focuses on ARIMA models.

AutoregressiveIntegratedMovingAverage(ARIMA)modelcanpredictthefuture price from historical data (time series datasets).The model uses no other independentvariablesbutthepredictionwillcomeoutfromhistoricalcocoaprices. So, it does not provide any theoretical backgrounds. ARIMA model requires large runoftimeseriesdataandtechnicalexpertiseonthepartofforecaster.Webelieved that it is reasonable to utilize ARIMA model to forecast cocoa prices with such a large sample size of 157 monthly observations from January 2000 to May 2014. Meyler et al. (1998) listed some advantages of using ARIMA model. Firstly, ARIMA model only requires data from the time series and a large dataset is preferable.

1.5   Significance of the Study

Theempiricalresultsandfindingsfromthestudywouldbesignificanttoindustries, practitioners and policy makers such as the government, businesses and the general public as well as academics and researchers due to the following:

Firstly, identifying the optimal model based on the relative performance, will be very useful in the planning activities of the government, businesses and the public in general.

Secondly, the results from the study will benefit academia and researchers by contributing to existing literature by closing or elimination the gap in literature or information on the relative performance of Maximum Likelihood Estimation (MLE) and Conditional Least Squares Estimation (CLSE). It will also serve as a basis for further research for both academic researchers and industry practitioners.

1.6   Organisation of the Study

This study has been arranged into five chapters. The first chapter constitutes the background of the study,statement of the problem,aims and objectives of the study, justification, methodology, significant of the study, and the organization of the study. Chapter 2 covers literature review on the theoretical and empirical works of previous researches. Chapter 3 covers the research methodology which deals with themethodstheresearchwilluseinachievingitstargetobjective.TheDataanalysis and simulation study from the international market monthly prices of cocoa are discussed in chapter 4. Chapter 5, which is the conclusion part, contains summary of salient points, conclusion of the research findings, suggestions and recommendations of the research work.

DEVELOPMENT OF AGRICULTURAL PRODUCE SALES FORECASTING SYSTEM: A CASE STUDY OF COCOA PRICES